RW-A-04: The Arusha Accords — Power-Sharing Negotiations, the Broad-Based Transitional Government, and the Architecture of Failure (1990–1994)

Status: DRAFTWords: 15,301

1. Key Takeaways

The Arusha Peace Agreement signed at Arusha, Tanzania, on 4 August 1993 by President Juvénal Habyarimana (for the Government of the Republic of Rwanda) and Colonel Alexis Kanyarengwe (Chairman of the Rwandan Patriotic Front, the displaced Hutu former Habyarimana ally who had defected to the RPF in the late 1980s and served as its political face), with the witness signatures of the regional heads of state and the OAU and UN representatives, was the negotiated terminus of the four-phase civil war that had run intermittently from the 1 October 1990 RPF invasion. It comprised the N'Sele Ceasefire Agreement of 29 March 1991 (later updated at Gbadolite, Bujumbura, and Kampala), and five substantive Protocols covering: (i) the Rule of Law (18 August 1992); (ii) Power-Sharing within the framework of a Broad-Based Transitional Government, Part I (30 October 1992); (iii) Power-Sharing Part II completing the institutional architecture (9 January 1993); (iv) the Repatriation of Refugees and Resettlement of Displaced Persons (9 June 1993); and (v) the Integration of the Armed Forces (3 August 1993). The Agreement is, alongside the Mozambican General Peace Accord (Rome, 4 October 1992) and the South African transitional arrangements (1993), the most institutionally elaborate African civil-war peace settlement of its decade — and the most catastrophically failed, in the eight months between signature and the 6 April 1994 shoot-down that terminated it.
The N'Sele Ceasefire of 29 March 1991, mediated by Zairian President Mobutu Sese Seko at his N'Sele residence near Kinshasa with subsequent updating at Gbadolite (16 September 1991) and Bujumbura (Bujumbura-1, 14 February 1991 and Bujumbura-2, March 1992 [TBD-VERIFY: precise Bujumbura date sequence is variably reported across sources, with the 14 February 1991 modus operandi and the subsequent March 1992 renewals]), and the Arusha I cessation-of-hostilities agreement of 12 July 1992, established the diplomatic infrastructure of the negotiation. They moved the mediation framework from the initially Zairian-led OAU process toward the Tanzanian-led OAU process that would carry the substantive protocols. The Tanzanian Ministry of Foreign Affairs (Foreign Minister Ahmed Hassan Diria and, in the closing phase, Foreign Minister Joseph Rwegasira) provided the principal mediation; OAU Secretary-General Salim Ahmed Salim (himself Tanzanian) supplied OAU institutional cover. Observer-state representation came from Belgium, France, Germany, the United States, the Holy See, and the UN — with the United States (the observer team led at various points by Assistant Secretary of State Herman Cohen through 1993 and Ambassador David Rawson subsequently) emerging as the most diplomatically engaged Western observer, and France the most politically conflicted given its concurrent military support to the FAR under Operation Noroît.
The 18 August 1992 Protocol on the Rule of Law committed both parties to a multi-party democratic order, fundamental rights guarantees, an independent judiciary, and the renunciation of any "exclusivist" ideology. It is the protocol that subsequent commentators have most often described as rhetorically successful and practically inert: the regime had already, under the 10 June 1991 multiparty constitution, formally introduced legal pluralism, and the protocol's commitments were largely declarative. Its content was the implicit prohibition on the Hutu Power ideology then crystallising in the Coalition pour la Défense de la République (CDR) and in the Kangura and (from July 1993) Radio Télévision Libre des Mille Collines (RTLM) propaganda apparatus — a prohibition that was never enforced, and whose non-enforcement is one of the principal documented failures of the Arusha process. Bruce Jones (2001) reads the Rule of Law protocol as exemplifying the broader weakness of Arusha: it produced agreement on text without producing agreement on the underlying political legitimation of pluralism that the text required.
The 30 October 1992 Protocol on Power-Sharing (Part I) and the 9 January 1993 Protocol on Power-Sharing (Part II) together produced the Broad-Based Transitional Government (BBTG) framework: a 21-member Council of Ministers with portfolio allocations to be filled across the existing recognised parties (MRND, MDR, PSD, PL, PDC) and the RPF, a 70-member Transitional National Assembly (TNA) with seats apportioned across the same parties plus smaller participants, a transitional President (Habyarimana retaining the role pending elections) with substantially reduced powers, and a transitional Prime Minister drawn from the MDR opposition. The portfolio allocations ultimately specified (after January 1993 and refined in the subsequent implementation negotiations) were: MRND: 5 ministries; RPF: 5 ministries; MDR: 4 ministries (including the Prime Minister); PSD: 3 ministries; PL: 3 ministries; PDC: 1 ministry. Critically, the protocols excluded the CDR — the Coalition pour la Défense de la République, founded in March 1992 and the principal Hutu Power extremist party — from BBTG participation, on the RPF's insistence that the CDR's racially exclusivist programme was incompatible with the Rule of Law protocol. The CDR exclusion is widely identified in the historiography (Jones 2001; Prunier 1995; Guichaoua 2015) as one of the principal spoiler-generating features of the settlement: it left a significant Hutu Power constituency outside the transitional institutions and therefore with no stake in their preservation.
The 8 February 1993 RPF offensive — a major resumption of hostilities launched in the wake of January 1993 anti-Tutsi massacres in the northwest (the Gisenyi and Ruhengeri killings of late January to early February 1993, documented contemporaneously by the International Commission of Investigation on Human Rights Violations in Rwanda that reported in March 1993) — broke the formal ceasefire and brought the RPF within roughly 30 kilometres of Kigali before being halted. The offensive killed [TBD-VERIFY: civilian casualty figure variously estimated at several hundred to over a thousand for the offensive period itself, with substantial displacement of perhaps 800,000–1,000,000 people from the northern prefectures fleeing the RPF advance, the displacement figure being one of the more durable consequences of the offensive]. The military effect was to demonstrate RPF military superiority and to compress Habyarimana's negotiating position; the political effect was to harden Hutu Power sentiment by giving content to the ibyitso ("accomplice") propaganda about Tutsi civilians as RPF auxiliaries. The post-offensive Dar es Salaam ceasefire of 7 March 1993 reactivated the negotiation track on terms more favourable to the RPF than those of late 1992 — and is therefore the proximate cause of the 9 June 1993 Refugee Protocol and the 3 August 1993 Armed Forces Protocol that completed the Arusha architecture on terms the regime's hardliners regarded as intolerable.
The 3 August 1993 Protocol on the Integration of the Armed Forces is the most-discussed and most-bitterly-resisted of the Arusha protocols. It provided for a unified Rwandan national army at a total strength of 19,000 troops to be composed in a 60:40 ratio of FAR to RPF personnel, with a gendarmerie of 6,000 personnel at the same 60:40 ratio — but with command positions distributed at 50:50 between the two parties, including the rotation of the most senior posts. The 60:40 ratios at troop level were a substantial concession to the FAR, which had approximately 40,000 personnel at the war's height and whose demobilisation under the protocol would be massive; the 50:50 command-position rule, however, gave the RPF a share of operational control disproportionate to its troop strength, including the Chief of Staff of the Gendarmerie. For the FAR officer corps — and particularly for the akazu-aligned senior officers around Colonel Théoneste Bagosora — the command-integration provisions represented the loss of the military instrument of regime survival. Bagosora's reported reaction at Arusha — "Je vais préparer l'apocalypse" / "I am going to prepare the apocalypse" — is widely cited in the historiography (Des Forges 1999; Melvern 2004) [TBD-VERIFY: the verbatim quotation as recorded by Tanzanian and French diplomatic interlocutors and reported in subsequent ICTR proceedings; the sentiment is undisputed; the precise wording across sources varies between "I am returning to Kigali to prepare the apocalypse" and the shorter form].
The 9 June 1993 Protocol on the Repatriation of Refugees and Resettlement of Displaced Persons addressed the principal grievance of the RPF — the right of return for the 600,000–700,000 Tutsi refugees displaced by the 1959–1964 violence and their descendants (the "old caseload" refugees in Uganda, Burundi, Tanzania, and Zaire), and for the more recent caseload displaced by the 1990–1993 civil war. The protocol established a Commission on Repatriation under tripartite (Rwanda-UNHCR-host country) governance, time-limited recovery of "vacant lands", and a recognition that not all returnees could be restored to their pre-1959 properties given the demographic transformation of the intervening thirty-four years. The protocol is the resolution that the post-1994 RPF-led government would inherit and partially supersede: the post-genocide return was vastly larger and more chaotic than Arusha contemplated, but the legal-conceptual architecture of "right of return without restitutionary maximalism" carried over.
The post-signature non-implementation across August 1993 – April 1994 is the operational architecture of the failure. The transitional institutions were to be installed within thirty-seven days of signature — i.e., by approximately 10 September 1993. They were not. The principal sequence of obstruction included: the dispute over CDR participation in the TNA (Habyarimana's MRND insisted on including CDR; the RPF and the moderate parties refused); the splitting of the MDR, PL, and PSD into "moderate" and "Hutu Power" wings, with the Hutu Power wings claiming entitlement to share of the transitional posts that had been allocated to the parties as wholes; Habyarimana's repeated failure to convene the swearing-in ceremony across November 1993, January 1994, February 1994, and March 1994; and the 5 January 1994 swearing-in of Habyarimana as transitional president without the simultaneous swearing-in of the Prime Minister Faustin Twagiramungu, the TNA, or the Council of Ministers — a procedural manoeuvre widely read as installing the BBTG presidency without the BBTG. The corpus follows Jones (2001) and Guichaoua (2015) in treating the non-implementation as a deliberate strategy of the akazu-aligned faction rather than as a series of contingent procedural failures.
UNAMIR — the United Nations Assistance Mission for Rwanda, established by Security Council Resolution 872 of 5 October 1993 under Force Commander Lieutenant-General Roméo Dallaire (Canadian) with Chief of Staff Major Brent Beardsley, with an authorised strength of 2,548 military personnel and a Chapter VI peacekeeping mandate — was the principal external mechanism for guaranteeing the Arusha settlement. Its mandate, deliberately constrained in the Security Council negotiation by the United States (which under the post-Somalia PDD-25 framework was attempting to limit peacekeeping commitments) and the United Kingdom, was confined to ceasefire-monitoring, security-zone supervision in Kigali, and "contribution to the security" of the transitional institutions; it lacked the Chapter VII authority and the rules of engagement to seize weapons caches or to interdict the militias whose preparations Dallaire would document. The 11 January 1994 cable from Dallaire to UN DPKO — the so-called "genocide fax", reporting an informant ("Jean-Pierre", subsequently identified as the Interahamwe trainer Jean-Pierre Abubakar Turatsinze [TBD-VERIFY: precise identification across sources]) warning of arms caches and the preparation of Tutsi extermination lists, and requesting authorisation to seize the caches — and the DPKO's reply (signed by Iqbal Riza on behalf of Under-Secretary-General Kofi Annan) instructing Dallaire not to seize the caches and to share the informant's information with Habyarimana, is the most-cited documentary instance of the international warning ignored.
The 21 October 1993 assassination in neighbouring Burundi of the recently elected Hutu president Melchior Ndadaye — the first Hutu president of Burundi, in office only since 10 July 1993, killed in a coup attempt by elements of the Tutsi-dominated Burundian army, with subsequent civil-war violence killing tens of thousands across the country — was the single most consequential external shock to the Rwandan Arusha process between signature and collapse. The Ndadaye killing was seized upon by Hutu Power propaganda inside Rwanda — and particularly by RTLM, which had begun broadcasting on 8 July 1993, just before the signature — as definitive proof that power-sharing with Tutsi was fatal for Hutu. Within months the Burundi shock had hardened Hutu Power sentiment within the MDR (with the Faustin Twagiramungu moderate wing increasingly isolated from the Donat Murego / Frodouald Karamira Hutu Power wing) and the PL (with Justin Mugenzi's Hutu Power wing splitting from the Landoald Ndasingwa moderate wing). The corpus reads the Burundi shock as having transformed an already fragile Arusha implementation into a structurally improbable one within 100 days of signature.
The 5 April 1994 date — set in the regional summit at Dar es Salaam on that day as a final deadline for the swearing-in of the BBTG — passed without action; on the evening of 6 April 1994, returning from Dar es Salaam where the regional heads of state (Tanzania's Ali Hassan Mwinyi, Uganda's Yoweri Museveni, Burundi's Cyprien Ntaryamira, Zaire's Faustin Birindwa as Mobutu's proxy [TBD-VERIFY: Zairian representation], and Kenya's Daniel arap Moi as observer) had pressed Habyarimana for implementation, his presidential Falcon 50 (registration 9XR-NN) was shot down on its approach to Kigali at approximately 20:25 local time. The genocide began that night with the killings of the moderate opposition, the Belgian peacekeepers, and the first Tutsi victims at roadblocks. The Arusha process — eight months and four days after its signature — was operationally terminated; the Arusha text, however, would survive as the legal-conceptual foundation of the post-genocide transitional arrangements documented in RW-A-01 and RW-C-05. The 1994 genocide's relationship to the Arusha process is captured in two reciprocal propositions: the genocide would not have occurred in the form it took without the Arusha process having raised the stakes of regime survival to existential levels; and the Arusha process could not have prevented the genocide once the akazu and Hutu Power factions had concluded that implementation was unsurvivable.
The corpus records three accounts of the Arusha failure which it does not adjudicate. The MRND-and-Habyarimana account, surviving partially in some Francophone polemical and revisionist literature, holds that the Arusha terms — particularly the 50:50 command-integration provisions and the CDR exclusion — were so politically unsurvivable that the regime could not have implemented them; the failure was structural and the RPF and the international mediators bear substantial responsibility for negotiating an unimplementable settlement. The RPF-and-Kagame account holds that Arusha was a serious good-faith settlement that the akazu faction sabotaged through a deliberate strategy of obstruction across August 1993 – April 1994 while preparing the genocide, and that the shoot-down was the final step in that strategy; the failure was a conspiracy. The international-mediator-and-Dallaire account, represented by Jones (2001), Adelman and Suhrke (1999), Dallaire (2003), and the Carlsson Report (1999), reads the failure as a combination of structural fragility (the CDR exclusion, the speed of the negotiated transition, the inadequate post-signature security architecture) and contingent diplomatic failure (the UNAMIR mandate's narrowness, the Belgian-and-American refusal to expand it after the January 1994 warnings, the Secretariat's reading of post-Somalia constraints), without resolving the question of whether Arusha was always doomed or was sabotageable but salvageable. The corpus follows the third reading in declining to adjudicate between the first two while documenting the third's internal heterogeneity.

2. The 1 October 1990 RPF Invasion and the Four-Phase Civil War (1990–1992)

2.1 The Invasion

The Arusha negotiations cannot be understood without the military balance they sought to terminate, and that balance was set by the RPF invasion of 1 October 1990. The Rwandan Patriotic Front — founded as a political-military movement in Kampala in December 1987 from the older Rwandan Alliance for National Unity (RANU) within the second-generation Tutsi-refugee community in Uganda — launched a conventional military invasion across the Kagitumba border post in northeastern Rwanda with approximately 2,500–4,000 fighters drawn from veterans of the Ugandan National Resistance Army (NRA) that had brought Yoweri Museveni to power in 1986. The senior commanders included Major-General Fred Rwigyema (commander, also Deputy Commander of the Ugandan NRA and Deputy Minister of Defence in Museveni's government), Major Paul Kagame (then absent at the US Army Command and General Staff College at Fort Leavenworth, returning to assume command after Rwigyema's death), Major Peter Bayingana, and Major Chris Bunyenyezi. The invasion was the institutional culmination of a Tutsi-refugee return movement that had failed repeatedly across 1961–1967 (the Inyenzi incursions documented in RW-A-03) and that had been preserved in the Ugandan diaspora through three decades of statelessness.

The initial invasion went badly. Rwigyema was killed on 2 October 1990 [TBD-VERIFY: the precise circumstances of Rwigyema's death — whether by a stray bullet from Rwandan government forces, by RPF internal-factional fire from Bayingana and Bunyenyezi, or by another cause — remain contested; the latter thesis was advanced in Ugandan and some Rwandan oppositional accounts and is the basis for the subsequent execution of Bayingana and Bunyenyezi at Kagame's order in late 1990]. The RPF lost roughly half its initial strength in the first ten days of operations against a FAR reinforced by Belgian paratroopers (deployed briefly in early October to evacuate European nationals), a Zairian contingent of approximately 500 troops dispatched by Mobutu (operationally ineffective and quickly withdrawn under accusations of indiscipline), and the more substantial French Operation Noroît force of approximately 300 troops initially, expanded later, deployed under President Mitterrand's Françafrique security framework. By late October the FAR had pushed the residual RPF force back across the Ugandan border. The conventional invasion having failed, Kagame returned from Fort Leavenworth in late October 1990, took command, executed Bayingana and Bunyenyezi for the alleged factional dissent that had cost Rwigyema's life, and reconstituted the RPF as a guerilla force operating from the Virunga Mountains along the northern border. The transformation from a conventional Ugandan-NRA-style force to a sustained insurgency took the winter of 1990–1991 to complete.

2.2 The Domestic Reaction Inside Rwanda

The invasion's domestic political effect inside Rwanda was transformative. On the night of 4–5 October 1990, the Habyarimana government staged a fake attack on Kigali — the so-called "phantom attack" in which gunfire was reported across the capital but no RPF forces were present — to justify the subsequent mass arrest of approximately 8,000–10,000 suspected RPF ibyitso (accomplices), the overwhelming majority of them Tutsi civilians, who were held without charge in stadium-and-prison detention centres for weeks; an estimated [TBD-VERIFY: dozens to hundreds] died of overcrowding and abuse. The arrests were the regime's first use of the ibyitso propaganda category that would dominate the period through to 1994. In parallel, the first round of anti-Tutsi massacres of the civil-war period began: the Kibilira massacres of 11–13 October 1990 in Gisenyi prefecture (approximately 350–400 killed, mostly Tutsi); the Bagogwe massacres of January–March 1991 in the northwest, targeting the Bagogwe Tutsi pastoral subgroup (approximately 300–1,000 killed across multiple sites [TBD-VERIFY: precise toll varies across sources]); and the Bugesera massacres of 4–9 March 1992 in southern Rwanda (approximately 277–300 killed). These episodes were documented contemporaneously by the International Commission of Investigation on Human Rights Violations in Rwanda — a joint Africa Watch, FIDH, and ICHRDD mission with the participation of African and European human-rights organisations — which conducted field investigations in January 1993 and published its findings in March 1993. The Commission's report was the first internationally-credible documentation that the regime was organising ethnic violence; its rapporteur the Senegalese jurist Bacre Waly Ndiaye subsequently filed a UN report in August 1993 that used the word genocide to characterise the trajectory.

2.3 The Four Phases of the War

The civil war is conventionally divided into four phases, each terminating in a ceasefire arrangement that approximated to the negotiating stages of Arusha. The first phase (October 1990 – March 1991) was the failed conventional invasion and the FAR's restoration of pre-invasion lines, terminating in the 29 March 1991 N'Sele Ceasefire. The second phase (April 1991 – January 1992) was the period of the reconstituted RPF insurgency from the Virunga base, characterised by low-intensity raids, FAR counter-insurgency in the northern prefectures, and the gradual political reform inside Rwanda (the 10 June 1991 multiparty constitution, the legalisation of opposition parties through 1991–1992) under combined RPF military pressure and donor demands. The third phase (mid-1992 – February 1993) opened with the Arusha I cessation of hostilities of 12 July 1992 that froze the lines and inaugurated the protocol negotiations; it terminated with the 8 February 1993 RPF offensive. The fourth phase (February – August 1993) comprised the offensive, the 7 March 1993 Dar es Salaam ceasefire, the negotiation of the final three protocols, and the 4 August 1993 signature. The four-phase pattern is the standard analytical division in Jones (2001) and Adelman and Suhrke (1999); Reyntjens (1994) uses a slightly different periodisation but the content is comparable.

2.4 The Military Balance at the Negotiating Table

By mid-1993 the military balance had shifted decisively. The FAR had grown from approximately 5,000 personnel in October 1990 to a wartime strength of approximately 30,000–40,000, with significant absorption of new recruits whose training was rudimentary and whose ethnic-political indoctrination was substantial; new units (the Réseau Zéro, the Garde Présidentielle's expansion, the Para-Commando Battalion) had been raised and trained with French assistance. The RPF had grown from roughly 2,000 fighters at the end of the failed invasion to approximately 15,000–20,000 by mid-1993, with significantly higher operational discipline, more recent combat experience, and a force-multiplier effect from Kagame's command. The 8 February 1993 offensive had demonstrated that, in conventional engagement, the RPF could push to within roughly 30 kilometres of Kigali against the largest concentration of FAR resources. This military balance — the RPF outnumbered roughly 2:1 by the FAR but operationally superior — is the proximate cause of the eventual 60:40 troop-integration / 50:50 command-integration ratios in the 3 August 1993 Armed Forces protocol: the FAR's numerical superiority justified the 60:40, while the RPF's operational quality justified the 50:50 command share. Neither side was negotiating from a position of military pre-eminence; both were negotiating from positions of mutual military exhaustion and external pressure.

3. From N'Sele to Arusha I: The Ceasefire Architecture (March 1991 – July 1992)

3.1 The N'Sele Ceasefire (29 March 1991)

The first formal ceasefire of the civil war was concluded at Mobutu Sese Seko's N'Sele residential complex outside Kinshasa on 29 March 1991, with mediation conducted under the auspices of the OAU and the participation of Zaire, Burundi, Uganda, Tanzania, and Belgium. The agreement was thin: a cessation of hostilities, the establishment of a Neutral Military Observer Group (NMOG) under OAU command [TBD-VERIFY: NMOG initial deployment dates, approximately mid-1991], and a commitment to subsequent political negotiations. The N'Sele text was the institutional precedent that subsequent Arusha agreements would amend rather than supersede: the 29 March 1991 N'Sele Ceasefire as updated at Gbadolite (16 September 1991), Bujumbura (Bujumbura-1, February 1991; Bujumbura-2, March 1992 — the precise date sequence is variably reported across sources, with the principal updating reportedly occurring in February 1991 to address the formal status of the NMOG deployment and again in March 1992 to renew the cessation in advance of negotiations) and finally at Arusha I, 12 July 1992, is the ceasefire whose successive amendments traced the diplomatic centre of gravity of the mediation from Mobutu's personal patronage to the OAU institutional framework under Tanzanian leadership.

The transition from Mobutu's mediation to the Tanzanian-OAU mediation was the single most consequential diplomatic transition of the pre-negotiation. Mobutu's mediation was politically untenable for the RPF: Zaire was the principal regional ally of Habyarimana, the Zairian contingent had fought alongside the FAR in the first weeks of the invasion, and Mobutu's personal-patronage style was unsuited to the institutional precision the protocols would require. The Tanzanian-OAU framework, by contrast, was acceptable to both parties: Tanzania under President Ali Hassan Mwinyi had a long history of neutral mediation in the Great Lakes (it had hosted Burundian and Ugandan refugee populations across the 1970s–1980s), the OAU Secretariat under Salim Ahmed Salim was Tanzanian-led, and Tanzania had no significant economic or strategic stake in the Rwandan outcome. The transition was effectively completed at the Dar es Salaam summit of 19 February 1991 at which regional heads of state recommended a Tanzanian-led OAU process [TBD-VERIFY: precise summit date and resolution text].

3.2 The 12 July 1992 Arusha I Cessation of Hostilities

The Arusha I cessation of hostilities of 12 July 1992 is the agreement that opened the negotiation phase. It moved the formal mediation venue from N'Sele/Bujumbura to the Arusha International Conference Centre in northern Tanzania (the same venue that would, two years later, host the International Criminal Tribunal for Rwanda established by UN Security Council Resolution 955 of 8 November 1994), and it established a more elaborate ceasefire verification architecture under the expanded NMOG II. The agreement was negotiated in the wake of the substantial political opening inside Rwanda — the April 1992 transitional government under Prime Minister Dismas Nsengiyaremye (MDR, a southern-Hutu moderate from Gitarama) which brought the opposition parties MDR, PSD, PL, and PDC into government for the first time since 1973, with significant cabinet representation for them (MDR with 3 ministries including the Prime Ministership, PSD 3, PL 3, PDC 1, MRND retaining 9 of the 19 portfolios). The MDR-PSD-PL-PDC opposition's entry into government changed the negotiating arithmetic: the Rwandan delegation at Arusha from July 1992 onward was not the MRND-Habyarimana monolith but a transitional government whose Foreign Minister Boniface Ngulinzira (MDR) was institutionally aligned with the opposition's preference for accommodation with the RPF.

The Rwandan delegation across the negotiations was led principally by Foreign Minister Ngulinzira and by Ambassador Jacques Bihozagara [TBD-VERIFY: Bihozagara is more commonly identified as an RPF delegate; the Rwandan-government delegation was led at various stages by Ngulinzira, by Defence Minister James Gasana, and by other figures, with the precise leadership across the protocols varying]. The RPF delegation was led principally by Patrick Mazimhaka and Pasteur Bizimungu (the Hutu former Habyarimana official who had defected to the RPF and would become the post-genocide president), with Tito Rutaremara in legal-drafting roles. The Tanzanian mediation was led by Foreign Minister Ahmed Hassan Diria (and from late 1993 by Joseph Rwegasira) with the senior diplomat Ami Mpungwe in the day-to-day mediation role. The OAU representation was supplied by Mohamed Sahnoun and others; UN representation by Macaire Pedanou. The observer states — Belgium, France, Germany, the United States, the Holy See — sent diplomatic observers with varying engagement.

3.3 The Negotiating Framework

Arusha I established the conceptual division of the negotiation into the five protocol areas subsequently formalised: rule of law, power-sharing (in two parts), refugee return, and armed-forces integration. The negotiation pattern across the year from July 1992 to August 1993 was distinctive: each protocol was negotiated as a discrete textual document, signed separately, and then integrated into the final 4 August 1993 Peace Agreement that incorporated the protocols by reference. This protocol-by-protocol approach — modelled in part on the Camp David Accords and the OAU's earlier mediation of the Eritrea-Ethiopia and other conflicts — had two consequences. It produced a documentary record of substantial textual precision: the Arusha Accords' final text runs to approximately 80 pages with the integrated protocols, and is among the most institutionally elaborate African civil-war settlements ever negotiated. But it also produced a settlement whose discrete components could be selectively obstructed: by the time the August 1993 final protocol was signed, the rule-of-law and refugee provisions had been on the books for a year and three months without implementation steps, and the regime had had ample time to develop obstruction strategies.

The negotiating cadence was demanding. The Rwandan and RPF delegations met at Arusha across approximately eleven discrete sessions between July 1992 and August 1993, with intervening recesses during which the delegations returned to Kigali and Mulindi (the RPF's headquarters in northern Rwanda) for consultations. The Tanzanian mediation supplied the institutional infrastructure — translation services, drafting support, conference logistics — and the OAU and UN supplied the political cover. The total negotiation time of approximately thirteen months is comparable to other major civil-war peace processes of the period (the Mozambican Rome negotiation ran from July 1990 to October 1992, approximately twenty-seven months; the South African Multi-Party Negotiating Process ran from May 1993 to December 1993, approximately seven months for the interim constitution).

4.1 The Rule of Law Protocol (18 August 1992)

The Protocol of Agreement Between the Government of the Republic of Rwanda and the Rwandese Patriotic Front on the Rule of Law, signed at Arusha on 18 August 1992, was the first protocol of the negotiation. Its core provisions, distributed across approximately fifteen articles, committed both parties to: (i) the establishment of a multi-party democratic system, with the explicit recognition of "national unity, democracy, pluralism, and respect for human rights"; (ii) fundamental rights guarantees including freedom of expression, association, and movement, drawn from the 1948 Universal Declaration and the 1981 African Charter on Human and Peoples' Rights; (iii) an independent judiciary, with provisions for the constitutional court and the supreme court that subsequent protocols would integrate into the BBTG institutional architecture; (iv) the rejection of any "exclusivist" ideology, with explicit reference to "all forms of discrimination, particularly that based on ethnicity, region, gender, clan, religion, or any other criterion"; and (v) provisions for political-prisoner release and for press freedom that would, in principle, have constrained the regime's Hutu Power propaganda apparatus.

The Rule of Law protocol's textual achievement was substantial; its operational achievement was minimal. As Jones (2001, ch. 4) observes, the protocol was negotiated rapidly because both parties found textual agreement easy: the regime's MRND-MDR-PSD-PL-PDC government formally accepted the principles, and the RPF was the protocol's principal author and beneficiary. But the protocol contained no enforcement mechanism — no constitutional-court referral procedure, no international monitoring of compliance, no consequence for the licensing of incitement media. When RTLM was licensed in mid-1993 and began broadcasting on 8 July 1993, three weeks before the final-Arusha signature, the licensure was a formal violation of the Rule of Law protocol's exclusivist-ideology prohibition; no international mediator invoked the protocol against it, and the regime did not enforce it against the broadcaster. Similarly, the Kangura magazine's continuous publication of the "Hutu Ten Commandments" framework from December 1990 onward — a text explicitly inconsistent with the protocol's anti-discrimination provisions — was never challenged on Arusha grounds.

The Protocol on Power-Sharing within the Framework of a Broad-Based Transitional Government, signed at Arusha on 30 October 1992, was the first of the two power-sharing protocols and the institutional centrepiece of the negotiation. It established the architectural skeleton of the BBTG: a transitional President of the Republic (Habyarimana, with substantially reduced powers); a transitional Prime Minister to be drawn from the MDR opposition; a Council of Ministers of an as-yet-unspecified size; a Transitional National Assembly (TNA) of an as-yet-unspecified size; and a transitional period running for approximately 22 months from BBTG installation, culminating in multiparty elections. The protocol left for subsequent negotiation the specific portfolio allocations and seat distributions, which would be completed in the 9 January 1993 protocol.

The protocol's distinctive design choice — the choice that the historiography identifies as both its principal merit and its principal vulnerability — was the deep reduction of presidential power. Under the protocol, Habyarimana as transitional President would retain ceremonial functions (head of state, representation of Rwanda externally, the formal commander-in-chief role) but would lose executive control: the Prime Minister and the Council of Ministers would direct the government; the President's decree powers would be sharply circumscribed; ministerial appointments and removals would require Council approval. The design was modelled in part on the Mozambican settlement's reduction of the presidential role under the General Peace Accord and in part on the French Fifth-Republic Cohabitation precedent. The merit was that it made the settlement institutionally credible to the RPF, which would not have accepted a settlement leaving Habyarimana with the executive instruments by which he had governed since 1973. The vulnerability was that it made the settlement existentially threatening to Habyarimana and the akazu, who would lose the patronage instruments — ministerial appointments, parastatal placements, regional-quota implementation — through which the regime had been maintained.

4.3 The CDR Question

The Coalition pour la Défense de la République (CDR), founded in March 1992 under leaders Jean-Bosco Barayagwiza and Martin Bucyana, was a Hutu Power extremist party that operated as the radical-flank political wing of the akazu faction. Its founding charter — explicitly ethnically exclusivist in framing Rwandan citizenship as a Hutu majoritarian project — placed it in textual conflict with the Rule of Law protocol's prohibition on exclusivist ideology. The Power-Sharing I protocol of October 1992, which specified that "parties subscribing to the Rule of Law protocol" would be eligible for transitional-institution participation, was widely understood by both parties to exclude the CDR from the BBTG. The RPF was emphatic on this point: the CDR's manifest racial programme was incompatible with the settlement's foundational principle. The MRND, while not formally aligned with the CDR (the two parties had distinct memberships and competing claims to the Hutu Power constituency), was reluctant to acquiesce in CDR exclusion because the CDR's exclusion would deprive the regime of a useful flanking party in the transitional arrangements.

The October 1992 protocol thus deferred the CDR question to subsequent negotiation, and the deferral would prove decisive: the CDR's status would become the principal procedural obstacle to BBTG installation across late 1993 and early 1994. Jones (2001) and Guichaoua (2015) treat the CDR-exclusion question as the clearest example of the Arusha process's spoiler problem: a substantial Hutu Power political constituency, with armed wings (the Impuzamugambi militia, founded in 1992 as the CDR youth wing in parallel to the MRND's Interahamwe) and propaganda assets (the Kangura magazine, in which CDR figures were prominent), was placed structurally outside the transitional institutions and therefore had no stake in their preservation. The literature on spoiler problems in peace processes (Stedman 1997; Walter 2002) treats this configuration as among the most reliably destabilising: where a militarised political constituency is excluded from a settlement, the settlement is significantly more likely to fail.

4.4 The Northern-Killings Episode and the Approach to January 1993

Between the October 1992 Power-Sharing I protocol and the January 1993 Power-Sharing II protocol, the political environment inside Rwanda deteriorated sharply. In late January – early February 1993, a series of anti-Tutsi massacres in the northwestern prefectures of Gisenyi, Ruhengeri, and Kibuye killed approximately 300 Tutsi civilians [TBD-VERIFY: the contemporary commission documented approximately 300 killed; some subsequent estimates run higher, with the cumulative toll across the January–February 1993 episodes variously placed between 300 and 1,000]. The killings were the proximate trigger for the International Commission of Investigation that conducted its January 1993 fieldwork and reported in March 1993 — and for the RPF's 8 February 1993 offensive that broke the ceasefire on the ground that the regime was using the negotiation period to organise violence rather than to implement the protocols. The protocols thus far signed were therefore being negotiated against a background of escalating regime-organised ethnic violence — a fact the historiography identifies as one of the principal reasons the eventual settlement was so structurally fragile: the parties were negotiating peace while one of them was preparing for war.

5. The 9 January 1993 BBTG Protocol and the 8 February 1993 RPF Offensive

The Protocol on Power-Sharing Within the Framework of a Broad-Based Transitional Government (Part II), signed at Arusha on 9 January 1993, completed the institutional architecture of the BBTG by specifying the portfolio allocations, the TNA seat distribution, and the transitional-period sequencing. The protocol's principal content was as follows.

The Council of Ministers would comprise 21 portfolios, allocated: MRND 5 (including the Ministry of Defence, the Foreign Ministry, and the Ministry of Family and Women's Affairs); RPF 5 (including the Ministry of the Interior, a critical portfolio given the prefecture-and-commune administrative chain it controlled); MDR 4 (including the Prime Ministership and the Ministry of Foreign Affairs [TBD-VERIFY: precise portfolio distribution within the MDR allocation across sources varies; the principal MDR portfolios across the various subsequent implementation drafts included Prime Minister, Foreign Affairs, Primary and Secondary Education, and Information]); PSD 3 (including the Ministry of Finance, contested and significant); PL 3; and PDC 1. The MRND's reduction from the 9 portfolios it had held in the April 1992 transitional government to 5 portfolios in the BBTG represented the regime's substantial concession; the RPF's elevation from 0 portfolios to 5 represented its incorporation into the state. The combined non-MRND share (RPF 5 + MDR 4 + PSD 3 + PL 3 + PDC 1 = 16 of 21 portfolios) gave the opposition coalition a structural majority in the cabinet — a feature the akazu faction regarded as politically intolerable.

The Transitional National Assembly (TNA) would comprise 70 seats, allocated: MRND 11; RPF 11; MDR 11; PSD 11; PL 11; PDC 4; smaller parties (PSR, MRSD, PDI, UDPR, RTD) 11 collectively [TBD-VERIFY: the precise small-party allocations vary in implementation drafts; the formula was nominally one seat per recognised small party with adjustments]. The notable feature was the equal MRND-RPF-MDR-PSD-PL share of 11 seats each, which gave no single party a structural majority and required coalition-building for every legislative vote. The PDC's lower allocation (4 seats) and the small parties' aggregate share (approximately 11 seats) were the residual elements.

The transitional period was set at 22 months from BBTG installation, after which multiparty presidential and legislative elections would be held, supervised by an international commission and conducted under the new constitutional arrangements. The BBTG was therefore intended as a bridging institution between the wartime regime and a post-war democratic order — not a permanent settlement. Habyarimana's position as transitional President was unambiguously time-limited; his political future depended on his prospects in the post-transition elections, which most regional and Western observers regarded as unfavorable to his MRND given the ethnic-and-regional arithmetic.

5.2 The Northern-Killings Trigger and the 8 February Offensive

The 9 January protocol was signed against a deteriorating security environment. Across late January and early February 1993, the Gisenyi, Ruhengeri, and Kibuye anti-Tutsi massacres killed approximately 300 civilians; the violence was organised by local MRND and CDR officials and carried out by the Interahamwe and Impuzamugambi militias with significant participation by ordinary residents of the northwestern hills. The International Commission of Investigation — the joint mission of Africa Watch, FIDH, ICHRDD, and the Inter-African Union for Human Rights — conducted fieldwork in Rwanda from 7 to 21 January 1993, documenting the killings, the systematic regime involvement, and the militia-and-list architecture. Its report, released on 8 March 1993, was the first internationally-credible documentation that the Habyarimana regime was preparing systematic violence; the term crimes against humanity was used, and the subsequent UN report by Special Rapporteur Bacre Waly Ndiaye in August 1993 would explicitly use the term genocide in a prospective sense.

The RPF's response was military. On 8 February 1993, the RPF launched its largest offensive of the civil war, breaking out of its northern positions and advancing rapidly south. Within eight days the RPF had pushed to within roughly 30 kilometres of Kigali, taking the towns of Ruhengeri (which fell on 8 February 1993) and Byumba (which had fallen earlier in 1993 and was now consolidated), and creating a refugee outflow of approximately 800,000–1,000,000 people fleeing south from the northern prefectures into the central-government-controlled area around Kigali. [TBD-VERIFY: the displacement figure of approximately one million is widely cited; some sources give 600,000–800,000.] The military effect was to demonstrate that, in the absence of a negotiated settlement, the RPF could take Kigali in a matter of weeks against the FAR as then configured.

The offensive's diplomatic effect was to compress the negotiating timetable and to shift the leverage decisively toward the RPF. The 7 March 1993 Dar es Salaam ceasefire, negotiated under intensive regional pressure (with Tanzanian, Burundian, and Ugandan mediation and OAU and UN supervision), restored the cessation of hostilities on terms more favourable to the RPF: the RPF retained the territory it had taken (the northern Byumba salient and surrounding areas, comprising roughly 8 per cent of Rwandan territory); a buffer zone was established between the parties under NMOG II monitoring; and the protocols-in-negotiation (refugees, armed forces) would be completed under the post-offensive military realities.

The offensive's domestic political effect inside Rwanda was the opposite of its diplomatic effect: it hardened Hutu Power sentiment by giving content to the ibyitso propaganda. The massacres of January–February had been read by the MRND-CDR coalition as a justified response to RPF infiltration; the offensive was read as proof that the RPF was a foreign Tutsi army willing to displace one million Hutu civilians and to march on Kigali, and that Tutsi civilians inside Rwanda were therefore credibly to be regarded as a fifth column. The propaganda apparatus, which had been gathering capacity throughout 1992, now operated against this background. The contradictory effect of the offensive — increasing RPF diplomatic leverage while simultaneously increasing the domestic-political risk of the eventual settlement it would secure — is one of the central analytical puzzles of the period and a principal subject of Kuperman's (2001) counterfactual analysis.

5.3 The Spring 1993 Reconfiguration of Rwandan Politics

The aftermath of the February offensive produced a reconfiguration of the Rwandan domestic political landscape that would shape the rest of the year. The opposition parties — MDR, PSD, PL, and PDC — split internally between moderate factions willing to continue the negotiation with the RPF (the Faustin Twagiramungu wing of MDR, the Frédéric Nzamurambaho wing of PSD, the Landoald Ndasingwa wing of PL) and Hutu Power factions increasingly aligned with the MRND's hard line (the Donat Murego and subsequently Frodouald Karamira wing of MDR, the Justin Mugenzi wing of PL). The PSD's principal figure Félicien Gatabazi would be assassinated on 21 February 1994 [TBD-VERIFY: precise date; commonly given as 21 February 1994], an attack that triggered retaliatory killing of the CDR leader Martin Bucyana the same day. The internal splits would have decisive consequences for the BBTG implementation: each party's portfolio allocation in the 9 January protocol had been made on the assumption of a unified party; the splits enabled the Hutu Power wings to claim entitlement to share of the allocations, which the moderate wings disputed, and the disputes were used by the MRND to delay BBTG installation across late 1993 and early 1994.

The MRND itself underwent a parallel internal reconfiguration. The party's formal leadership remained with Habyarimana, but the operational direction of the Hutu Power faction passed increasingly to figures including Colonel Théoneste Bagosora (the Defence Ministry's directeur de cabinet), Captain Pascal Simbikangwa, and the akazu network around Madame Agathe Habyarimana. These figures organised the militia-arming, the propaganda apparatus, and (in the historiographical reading documented by Des Forges 1999 and the ICTR Bagosora judgement of 18 December 2008) the early operational preparations for the killing. The April 1992 transitional government's MDR Prime Minister Dismas Nsengiyaremye had been increasingly sidelined; in July 1993 he was replaced by Agathe Uwilingiyimana (also MDR, but of the moderate Twagiramungu wing) as Prime Minister, in a manoeuvre that displaced Nsengiyaremye and weakened the moderate position within the cabinet [TBD-VERIFY: the precise circumstances of Nsengiyaremye's replacement vary in detail across sources].

6. Refugees and Armed Forces: The June and August 1993 Protocols

6.1 The 9 June 1993 Protocol on Repatriation of Refugees

The Protocol of Agreement on the Repatriation of Rwandese Refugees and the Resettlement of Displaced Persons, signed at Arusha on 9 June 1993, addressed the principal grievance of the RPF: the right of return for the Tutsi refugees displaced by the 1959–1964 violence and their descendants. The "old caseload" comprised approximately 600,000–700,000 persons of Rwandan origin and their descendants resident in Uganda (the largest community, approximately 200,000–300,000), Burundi (approximately 150,000–200,000), Tanzania (approximately 40,000–60,000), and Zaire (approximately 80,000–100,000) [TBD-VERIFY: precise figures vary across UNHCR and academic sources; the cumulative estimate of approximately 600,000–700,000 is the most-cited]. The "new caseload" comprised the internally displaced and externally displaced of the 1990–1993 civil war, including the roughly one million displaced by the February 1993 offensive.

The protocol's provisions, distributed across approximately twenty articles and four chapters, included: (i) the affirmation of the inalienable right of return for all Rwandan refugees, with the recognition that "the choice of where to settle is a fundamental right of every Rwandese citizen"; (ii) a time limit on restitution claims for property — refugees who had been absent for less than ten years would have a presumptive right to recover their original property, while those absent for more than ten years (the overwhelming majority of the old caseload, displaced thirty-four years earlier) would have no automatic restitution right and would receive resettlement assistance in "vacant lands" or in new locations of their choice; (iii) the establishment of a Commission on Repatriation under tripartite (Rwandan Government, UNHCR, host country) governance with responsibilities for organising the return; (iv) provisions for education-and-civil-service quota suspension during the resettlement period; and (v) international financial support for the resettlement, to be coordinated through UNHCR.

The ten-year restitution-cutoff provision was the most distinctive design choice. The provision was negotiated specifically to address the situation of the long-displaced old caseload: in the thirty-four years since 1959, the properties of the displaced had been reallocated under Rwandan land tenure, often multiple times, and full restitution would have required the displacement of the current Hutu occupants, an outcome incompatible with peaceful resettlement. The "vacant lands" provision — directing the resettlement of long-displaced refugees to the eastern lowlands (the Akagera National Park and surrounding areas, then sparsely populated) and to other areas not currently under intensive cultivation — was the compromise. The provision would survive the genocide: the post-1994 RPF-led government largely followed the Arusha approach in its 1994–1996 management of the old-caseload return, with the substantial caveat that the post-genocide context (including the abandonment of property by génocidaires who had fled to Zaire) created restitutionary opportunities that the Arusha protocol had not contemplated. Susan Thomson (2013) treats the legal-conceptual continuity between the Arusha refugee protocol and the post-1994 resettlement framework as one of the durable institutional legacies of the negotiation.

6.2 The 3 August 1993 Protocol on Integration of the Armed Forces

The Protocol of Agreement on the Integration of the Armed Forces of the Two Parties, signed at Arusha on 3 August 1993 — one day before the umbrella Peace Agreement signature — is the most-discussed and most-consequential of the protocols. It established the unified post-transitional Rwandan National Army and the unified National Gendarmerie, on the following terms.

The National Army would have a total strength of 19,000 troops, composed in a 60:40 ratio of FAR to RPF personnel — i.e., 11,400 from the FAR (a substantial demobilisation from the wartime FAR strength of approximately 30,000–40,000) and 7,600 from the RPF (a near-total incorporation of the RPF military force at the war's end, estimated at approximately 15,000–20,000, though with significant flexibility in the integration timeline). The National Gendarmerie would have a total strength of 6,000 personnel, similarly composed in 60:40 ratio (3,600 FAR, 2,400 RPF). The combined post-integration security force of 25,000 personnel represented a substantial overall reduction from the wartime combined total of approximately 45,000–55,000, with the demobilised personnel of both forces to be reabsorbed into civilian life through a Demobilization Programme coordinated with UNDP and other international donors.

The command-level provisions were the negotiating innovation. Command positions in both the army and gendarmerie were distributed at 50:50 between the two parties — with critical posts rotated between FAR and RPF officers. The Chief of Staff of the Army would be drawn from the FAR; the Deputy Chief of Staff from the RPF (or vice versa under some implementation drafts). The Chief of Staff of the Gendarmerie would be drawn from the RPF [TBD-VERIFY: the precise initial-rotation provisions across drafts vary; the principal substance is the 50:50 command share]. Battalion commands were similarly distributed 50:50, with the two battalions stationed in Kigali to be commanded one by each party. The RPF was authorised to deploy a 600-strong battalion to Kigali in advance of full integration, to be stationed at the CND (Conseil National de Développement), the Rwandan parliament building, under UNAMIR supervision — a provision implemented in December 1993 that would prove militarily decisive in April 1994, when the CND battalion would provide the initial RPF military presence at the centre of the capital from which the broader resumption of operations could be coordinated.

The disparity between the 60:40 troop ratio and the 50:50 command ratio was the protocol's most politically loaded feature. From the FAR officer corps' perspective, the 50:50 command share was disproportionate to the RPF's troop share and would give RPF officers operational control over substantial parts of the unified force. From the RPF's perspective, the 60:40 troop share was a substantial concession given its operational superiority. The compromise reflected the underlying military balance — the FAR numerically superior, the RPF operationally superior — and was the most that either delegation believed its principals would accept. Colonel Théoneste Bagosora, the senior FAR figure in the negotiating delegation, rejected the formula and is widely reported to have departed Arusha in protest before the protocol's signature [TBD-VERIFY: the timing and circumstances of Bagosora's departure across sources; the principal claim of his statement — "Je vais préparer l'apocalypse" / "I am going to prepare the apocalypse" — is widely cited in subsequent secondary literature including Des Forges (1999) and Melvern (2004), with attribution typically to Tanzanian diplomatic interlocutors; the verbatim quotation has not been verified against a primary documentary source and should be treated as paraphrase].

7. The 4 August 1993 Signature and the Spoiler Dynamics: CDR, RTLM, and Hutu Power

7.1 The Signature Ceremony

The Arusha Peace Agreement was signed at the Arusha International Conference Centre on 4 August 1993 in a ceremony witnessed by the regional heads of state — Tanzanian President Ali Hassan Mwinyi (the principal mediating head of state), Ugandan President Yoweri Museveni, Burundian President Melchior Ndadaye (Hutu, recently elected in June 1993, who would be assassinated within eleven weeks), Zairian Foreign Minister or proxy representing Mobutu Sese Seko [TBD-VERIFY: precise Zairian representation], and OAU Secretary-General Salim Ahmed Salim. Habyarimana signed for the Rwandan Government; Colonel Alexis Kanyarengwe signed for the RPF (Kagame as Vice-Chairman did not sign, the chairmanship role being formally Kanyarengwe's). The witness signatures of the UN, OAU, and observer states were affixed. The Peace Agreement text — running to approximately 80 pages with the five protocols integrated — was deposited with the Tanzanian Government and circulated as a UN document.

The signature ceremony was, by participant accounts, outwardly celebratory and inwardly unconvinced. The Tanzanian and OAU mediators were satisfied with the textual achievement; Western diplomatic observers, particularly Belgian and American, expressed reservations about the implementation prospects in private channels recorded subsequently. Habyarimana's body language was described in contemporaneous reporting as resigned rather than committed; Bagosora's pre-signature departure had been noted. The RPF delegation departed Arusha with the settlement they had sought; the senior FAR delegates departed with a settlement many of them regarded as politically and militarily unsurvivable for the regime.

7.2 The CDR-RTLM-Hutu Power Spoiler Architecture

In the eight months between signature and collapse, the spoiler architecture consolidated and developed. Its principal elements were the following.

The Coalition pour la Défense de la République (CDR) intensified its political mobilisation against the settlement. Excluded from the BBTG by the Rule of Law protocol's exclusivist-ideology prohibition and the consequent RPF veto, the CDR organised street-level violence and parliamentary obstruction. Its militia, the Impuzamugambi ("those with a single purpose"), was expanded and armed alongside the larger MRND Interahamwe ("those who attack together"); together the two militias by late 1993 numbered an estimated 30,000–50,000 trained fighters distributed across the prefectures [TBD-VERIFY: the militia-strength figures vary across sources, with lower estimates around 15,000 and higher around 50,000 across the Interahamwe-Impuzamugambi combination]. Militia training was conducted at multiple sites including military camps; weapons distribution to communes proceeded under cover of the constitutional civil-defence framework but with a distribution profile aligned with the Hutu Power mobilisation. The ICTR Bagosora judgement (18 December 2008) and Media Trial judgement (3 December 2003) document substantial parts of this preparation.

Radio Télévision Libre des Mille Collines (RTLM) began broadcasting on 8 July 1993 — twenty-seven days before the Arusha signature, and shortly after its founding by a consortium of MRND-CDR-aligned shareholders including Ferdinand Nahimana, Jean-Bosco Barayagwiza, Felicien Kabuga, and others. The station's broadcasting profile was distinctive: youth-oriented music programming alongside political commentary, the political content escalating in its ethnic-extremist framing across late 1993 and early 1994. The RTLM operation was a textual violation of the Arusha Rule of Law protocol; no enforcement followed. The station's pre-genocide broadcasts increasingly identified Tutsi civilians as RPF auxiliaries, named specific individuals as targets, and modelled the propaganda discourse that would, after 6 April 1994, direct the killing in real time. The ICTR Media Trial judgement convicted Nahimana, Barayagwiza, and Ngeze for direct and public incitement to commit genocide.

The Hutu Power label — coined publicly at a rally addressed by Frodouald Karamira (MDR-Power) on 23 October 1993 [TBD-VERIFY: precise date of the Hutu Power rally varies in sources, often given as 23 October 1993 with some sources citing late October or November] — supplied the explicit political framing under which the moderate-Hutu-Power splits within the opposition parties operated. Karamira, by then a leader of the MDR-Power faction, called for a "Pawa" (Power) alliance across the Hutu factions of all parties — MDR-Power, PL-Power, PDC-Power — in coalition with the MRND and CDR against the RPF and the moderate opposition. The Hutu Power coalition that emerged through late 1993 and early 1994 was, in effect, the political vehicle of the Arusha repudiation: it offered a vision of post-Arusha politics in which the RPF was excluded from the transitional institutions and the Hutu Power parties governed without the moderate Hutu and without Tutsi participation. The coalition was the political base from which, after 6 April 1994, the interim government under President Théodore Sindikubwabo and Prime Minister Jean Kambanda would be drawn.

7.3 The Spoiler Problem in Theoretical Frame

The Arusha case is among the most-cited illustrations of the spoiler problem in peace-process scholarship. Stephen Stedman's foundational 1997 article ("Spoiler Problems in Peace Processes", International Security 22:2) defined spoilers as "leaders and parties who believe that peace emerging from negotiations threatens their power, worldview, and interests, and use violence to undermine attempts to achieve it"; Stedman's typology distinguished limited spoilers (seeking marginal concessions), greedy spoilers (seeking more substantial gains), and total spoilers (seeking the destruction of the peace process itself). The CDR and the Hutu Power faction within the MRND fit the total-spoiler category: their political programme was incompatible with any settlement involving RPF inclusion, and their preparation for violence proceeded in parallel with the formal negotiation. Stedman's framework prescribes three responses to total spoilers — coercion, inducement, or socialisation — none of which the Arusha process effectively deployed against the Hutu Power faction. The UNAMIR mandate was non-coercive; inducement (offers of post-transitional positions) was not extended to the CDR on rule-of-law grounds; socialisation through institutional incorporation was precisely what the protocol architecture excluded. The literature thus reads Arusha as a textbook case of a peace process whose protocol design generated rather than managed its principal spoiler.

8. UNAMIR Deployment, the Ndadaye Shock, and the Dallaire Cable (October 1993 – January 1994)

8.1 The UNAMIR Mandate Negotiation

The Arusha Peace Agreement contemplated the deployment of a Neutral International Force to oversee the implementation of the transitional arrangements. The agreement's references to the force assumed a substantial mandate including the supervision of the integration of the armed forces, the security of the transitional institutions, the demobilisation of surplus military personnel, and the provision of confidence-building security throughout the transitional period. The negotiation of the actual UN force mandate in New York during August–October 1993 produced a substantially narrower instrument.

The Reconnaissance Mission led by Brigadier-General Roméo Dallaire of the Canadian Forces, dispatched to Rwanda from 19 August to 31 August 1993, recommended a force of approximately 4,500 troops with a robust Chapter VI mandate including the seizure of weapons caches, the disarmament of irregular forces, and the active protection of the transitional institutions. The recommendation was substantially reduced in the subsequent Secretariat negotiations. The UN Department of Peacekeeping Operations (DPKO) — under Under-Secretary-General Kofi Annan, with Iqbal Riza as Annan's deputy and the principal Rwanda-file officer — produced the report S/26488 of 24 September 1993 that recommended a force of approximately 2,548 troops with a constrained mandate. The constraint reflected three pressures: the US PDD-25 framework then being drafted in response to the 3–4 October 1993 Black Hawk Down events in Mogadishu (the deaths of 18 US Army Rangers in the Battle of Mogadishu, with the consequence of substantial American restraint on subsequent peacekeeping deployments) [TBD-VERIFY: PDD-25 was formally promulgated 3 May 1994, after the UNAMIR deployment, but its drafting context was shaping US Security Council positions from September 1993 onward]; the Belgian government's preference for a force not so substantial that Belgium's contribution would become politically exposed; and the broader Security Council reluctance to expand peacekeeping budgets in the post-Cold-War expansion of UN demands.

Security Council Resolution 872 of 5 October 1993 formally authorised UNAMIR with a 2,548-troop authorisation, a Chapter VI mandate confined to ceasefire monitoring, security supervision in Kigali, the establishment of a weapons-secure zone, and "contribution to the security" of the transitional institutions. The mandate explicitly did not authorise the seizure of arms caches, the active disarmament of militias, or the protection of civilians from non-Arusha-related violence. Dallaire was appointed Force Commander; Jacques-Roger Booh-Booh of Cameroon was appointed Special Representative of the Secretary-General with overall political direction. The UNAMIR force began deploying in October–November 1993 and reached approximately 80 per cent of authorised strength by early 1994, with the principal national contingents Belgian (the KIBAT battalion, approximately 400 troops, the best-equipped contingent and the principal Kigali force), Bangladeshi (the largest contingent, approximately 940 troops), Ghanaian (approximately 800 troops), and Tunisian (a smaller contingent).

8.2 The Ndadaye Assassination and the Burundi Shock (October 1993)

The Arusha process suffered its single most consequential external shock on 21 October 1993 with the assassination in neighbouring Burundi of President Melchior Ndadaye. Ndadaye, leader of the FRODEBU (Front pour la Démocratie au Burundi) Hutu-majoritarian party, had won the 1 June 1993 Burundian presidential election with 64.8 per cent against the incumbent Tutsi president Pierre Buyoya, and had taken office on 10 July 1993 — three weeks before the Arusha signature — as Burundi's first elected Hutu president. His government, drawn from FRODEBU and including substantial Tutsi participation, represented the most serious attempt at Hutu-Tutsi power-sharing in Burundian history. On the night of 20–21 October 1993, elements of the Tutsi-dominated Burundian army staged a coup attempt, captured Ndadaye at the presidential palace, and killed him at the military camp. Six senior FRODEBU figures including the Speaker of the National Assembly Pontien Karibwami and the Foreign Minister-designate were also killed.

The Burundian civil war that followed killed an estimated 50,000–100,000 people across late 1993 and 1994 [TBD-VERIFY: precise figures across sources vary, with most estimates in the range of 50,000–100,000 for the October 1993 to mid-1994 period before the Burundian conflict's separate later trajectory] and produced a refugee outflow of approximately 300,000–700,000 Burundian Hutu into Rwanda and Tanzania. The Burundian Hutu refugees who arrived in Rwanda in late 1993 — primarily in the southern and eastern prefectures — brought with them direct testimony of Tutsi-organised mass killing, which RTLM and CDR propaganda amplified into the dominant Hutu Power narrative of late 1993 and early 1994: that power-sharing with Tutsi is fatal for Hutu, that the Arusha settlement would produce in Rwanda what FRODEBU's attempted power-sharing had produced in Burundi.

The political effect inside Rwanda was decisive. The Hutu Power coalition's formation, which had been incipient before October, crystallised in the weeks after Ndadaye's death. The 23 October 1993 Karamira rally at which the "Pawa" framework was articulated occurred two days after the Burundi assassination and was substantially organised in response to it. The MDR, PL, and PDC splits that had been latent before October became operational in the weeks after; the moderate wings were progressively isolated; the Hutu Power coalition's claim that Arusha was inherently suicidal acquired empirical content. Bruce Jones (2001) and André Guichaoua (2015) treat the Ndadaye assassination as the single most damaging external event to the Arusha implementation; Kuperman (2001) similarly identifies it as the proximate cause of the post-October 1993 hardening of Hutu Power mobilisation.

8.3 The Implementation Failure of Autumn 1993

Under the Arusha timetable, the transitional institutions were to be installed within 37 days of signature — i.e., by approximately 10 September 1993. They were not. The principal procedural obstacles were the following.

The CDR-TNA seat question: the MRND insisted that the CDR was entitled to a seat in the Transitional National Assembly, on the ground that the CDR was a registered political party in Rwanda and that the Rule of Law protocol's exclusivist-ideology prohibition was insufficient to exclude it; the RPF refused, on the ground that CDR participation would violate the foundational protocols. The dispute was framed as procedural but its political content was the larger question of whether the Hutu Power constituency would be inside or outside the transitional institutions.

The opposition-party-split question: as the MDR, PL, and PDC split between moderate and Hutu Power wings, the question of which wing was entitled to claim the party's portfolio and TNA allocations became contested. The MRND insisted that the Hutu Power wings of each party should share the allocations; the moderate wings (backed by the RPF) refused. The dispute was procedural in form but again concerned the Hutu Power coalition's structural penetration of the transitional institutions.

The RPF battalion deployment to Kigali: under the Armed Forces protocol, the RPF was authorised to deploy a 600-strong battalion to the CND in advance of full integration. The deployment occurred on 28 December 1993 under UNAMIR supervision, and the battalion was stationed at the parliament building with strict UNAMIR-monitored confinement rules. The deployment was operationally successful and would prove militarily critical in April 1994; politically, however, it was used by Hutu Power propaganda as evidence that the RPF had infiltrated the capital and was preparing a military takeover.

Across November 1993, December 1993, January 1994, February 1994, and March 1994, Habyarimana repeatedly scheduled the swearing-in of the transitional institutions and repeatedly failed to convene it. The most-discussed instance was the 5 January 1994 ceremony at which Habyarimana was sworn in as transitional President without the simultaneous swearing-in of the Prime Minister (Twagiramungu of MDR-moderate), the TNA, or the Council of Ministers — a procedural manoeuvre installing the BBTG presidency without the BBTG and widely read as a deliberate strategy of obstruction. The historiographical reading of these failures has converged on the akazu-strategy interpretation: Jones (2001), Guichaoua (2015), Des Forges (1999), and Melvern (2004) all treat the obstruction as a deliberate strategy of the Hutu Power faction to delay implementation while preparing the alternative of genocide.

8.4 The Dallaire Cable (11 January 1994)

The most-documented warning of the impending violence is the cable transmitted by Dallaire to UN DPKO on 11 January 1994 — the so-called "genocide fax". The cable, classified UNAMIR-CONFIDENTIAL and addressed to Major-General Maurice Baril (UN Military Adviser) with copies to DPKO, reported information received from an informant identified by the codename "Jean-Pierre" (subsequently widely reported as Jean-Pierre Abubakar Turatsinze, a senior trainer of the Interahamwe who had requested UNAMIR protection in return for his testimony [TBD-VERIFY: the precise identification of "Jean-Pierre" across sources varies]). The informant reported: (i) that the Interahamwe had compiled lists of Tutsi residents of Kigali to be killed at the time of an unspecified trigger; (ii) that arms caches had been distributed across Kigali for distribution to the militias; (iii) that the Interahamwe training programme could "kill up to 1,000 Tutsi in 20 minutes" in Kigali; and (iv) that the informant's recent assignment had included planning the killing of Belgian UNAMIR peacekeepers in order to provoke a Belgian withdrawal.

Dallaire's cable requested authorisation to seize the arms caches identified by the informant under UNAMIR's mandate and to extend protection to the informant and his family. The reply, transmitted on 11 January 1994 from DPKO over the signature of Iqbal Riza on behalf of Kofi Annan, denied authorisation to seize the caches, on the ground that such action exceeded UNAMIR's Chapter VI mandate; it instructed Dallaire instead to share the informant's information with Habyarimana and with the Belgian, French, and US ambassadors, and to seek the regime's own action against the militia preparations. The reply has been the most-debated DPKO instruction of the period; Annan's subsequent position (variously stated in his post-2000 reflections) has been that the constraint reflected the mandate as authorised by the Security Council and that any expansion would have required Council authorisation that was not politically obtainable. The Carlsson Report (1999) treated the Riza-Annan instruction as institutionally explicable within the post-Somalia Secretariat posture but a failure of the warning system.

The information from Jean-Pierre was duly shared with the regime — i.e., with the political leadership that was directing the preparations the informant was warning against. The informant subsequently went into hiding and disappeared from the record [TBD-VERIFY: the post-January 1994 fate of "Jean-Pierre" is variably reported, with most accounts indicating that he survived but lost his protection from UNAMIR]. The arms caches were not seized. The militia preparations continued and intensified.

9. The Stalemate of Early 1994 and the Collapse on 6 April

9.1 February – March 1994: The Killings Cycle

The first quarter of 1994 produced a cycle of high-profile political killings that progressively destroyed the political space for Arusha implementation. On 21 February 1994 the PSD leader Félicien Gatabazi was assassinated in Kigali; later the same day the CDR leader Martin Bucyana was killed in apparent retaliation in Butare. The reciprocal killings demonstrated that the political-elite security architecture had collapsed; subsequent assassinations and assassination attempts across late February and March, accompanied by riots and roadblocks in Kigali at which the Interahamwe practised the operations they would conduct in April, made the daily security situation untenable. UNAMIR's effectiveness was constrained by the mandate; Dallaire's repeated requests for rules-of-engagement expansion were declined.

9.2 The 5 April Deadline and the Dar es Salaam Summit

By late March the regional heads of state had concluded that Habyarimana's procedural obstruction was a deliberate strategy and that further delay was incompatible with the survival of the Arusha framework. A regional summit was convened at Dar es Salaam on 6 April 1994 [TBD-VERIFY: some accounts place the summit on 5 April and the return flight on 6 April; the principal accounts place the summit on 6 April 1994 with the return flight occurring later the same day], attended by Habyarimana, Burundian President Cyprien Ntaryamira (Ndadaye's successor, who would die alongside Habyarimana), Tanzanian President Ali Hassan Mwinyi, Ugandan President Yoweri Museveni, Zairian representation, and Kenyan President Daniel arap Moi. The summit pressed Habyarimana for definitive implementation of the BBTG with a deadline that, by widely-reported but variably-cited accounts, was set for 5 April 1994 or immediately upon Habyarimana's return. Habyarimana, by participant accounts, agreed at the summit to implement the BBTG upon his return to Kigali.

9.3 The 6 April Shoot-Down

Habyarimana boarded his French-supplied Dassault Falcon 50 (registration 9XR-NN) at Dar es Salaam in the early evening of 6 April 1994 for the return flight to Kigali. Ntaryamira and the Burundian delegation joined the flight. The aircraft approached Kanombe airport in Kigali at approximately 20:25 local time and was struck on its final approach by two surface-to-air missiles, crashing in the grounds of the presidential residence adjacent to the airport. All twelve persons aboard were killed: Habyarimana; Ntaryamira; Burundian Minister of Communications Bernard Ciza; Burundian Minister of Planning Cyriaque Simbizi; FAR Chief of Staff Major-General Déogratias Nsabimana; directeur de cabinet Colonel Elie Sagatwa; the ambassador Juvenal Renzaho [TBD-VERIFY: ambassadorial passenger detail varies]; and the three French crew Jacky Héraud, Jean-Pierre Minaberry, and Jean-Michel Perrine.

The identity of those who fired the missiles remains the single most-contested individual question in Rwandan historiography, and is treated at length in RW-B-01 and RW-H-PRES-01. The two principal theses are the Hutu-extremist-coup thesis (that akazu-aligned figures within the FAR shot down their own president to remove the Arusha-implementation obstacle and trigger the prepared genocide) and the RPF-responsibility thesis (advanced in the November 2006 Bruguière indictments and in some Francophone polemical literature). The Mutsinzi Report (April 2009, published January 2010), the Trévidic-Poux French judicial investigation (2010–2018, closed with a non-lieu in December 2018), and the broader academic literature have produced findings broadly supportive of the Kanombe-firing-point location (consistent with FAR control of the launch zone), without conclusively settling the question. The corpus position, consistent with RW-B-01, is that the question is genuinely unresolved as of 2026 and that the authorship does not change the legal-historical character of the genocide that followed.

9.4 The Operational Termination of Arusha

The Arusha process was operationally terminated within hours of the shoot-down. The interim government convened on 8–9 April 1994 under President Théodore Sindikubwabo (former president of the CND and a Hutu Power MRND figure) and Prime Minister Jean Kambanda (MDR-Power) drew its composition from the Hutu Power wings of the parties; the moderate Prime Minister Agathe Uwilingiyimana had been killed by the Garde Présidentielle on the morning of 7 April; the Constitutional Court president Joseph Kavaruganda and senior moderates of the MDR, PSD, and PL had been killed in the first hours; the ten Belgian UNAMIR peacekeepers guarding Uwilingiyimana had been killed at Camp Kigali on 7 April; the Belgian government announced its withdrawal from UNAMIR on 14 April; the Security Council reduced UNAMIR to 270 personnel under Resolution 912 of 21 April. The transitional institutions specified by Arusha — the BBTG Council of Ministers, the TNA, the unified armed forces — were never installed.

The RPF resumed offensive operations within hours of the shoot-down. The CND battalion under Colonel Charles Kayonga broke out of its confinement on the night of 6–7 April and established positions in northern Kigali; the main RPF force under Major-General Kagame moved south from the northern bases; the four-month campaign that followed would take Kigali on 4 July 1994 and end the genocide through battlefield defeat of the Hutu Power state. The Arusha framework, however, would survive textually: when the RPF declared the Government of National Unity on 19 July 1994 it claimed to govern under the Arusha Accords as modified by the post-genocide reality, with the Constitutional Court restored under Arusha provisions and the TNA reconstituted on Arusha-inflected lines but with the MRND and CDR excluded as parties responsible for the genocide. The post-genocide transitional arrangements documented in RW-A-01 and the future RW-C-05 were institutionally continuous with Arusha while superseding it.

10. Three Accounts of the Arusha Failure

The corpus records three accounts of the Arusha failure, in the discipline applied to genuinely contested questions throughout the Rwanda documents.

10.1 The MRND-and-Habyarimana Account

The first account, surviving in partial form in the post-1994 Hutu-diaspora and revisionist Francophone literature (and in some elements of the Bruguière indictments and the work of journalists such as Pierre Péan), holds that the Arusha settlement was structurally unimplementable on its own terms. Three claims constitute this account.

First, the integration provisions of the 3 August 1993 Armed Forces protocol — the 50:50 command share against the 60:40 troop share — produced a unified force in which RPF officers would have operational control over substantial parts of the security architecture, an outcome the FAR officer corps could not be expected to accept and which would have created in the post-transition Rwanda a security architecture controlled by the militarily superior RPF regardless of the formal portfolio allocations. The Bagosora reaction at Arusha, on this account, was a rational response to a settlement that the FAR could not survive.

Second, the portfolio allocations of the 9 January 1993 Power-Sharing II protocol — the 16-of-21 non-MRND share — produced a transitional cabinet in which the MRND was a structural minority and would, given the RPF-MDR-PSD-PL coalition that the allocations enabled, be unable to influence the post-transition trajectory. The MRND's reduction from 9 to 5 portfolios and the RPF's elevation from 0 to 5 reversed the political arithmetic in a way the regime's surviving institutional base could not accept.

Third, the CDR exclusion removed a significant Hutu Power constituency from the institutional bargain while leaving its militia and propaganda assets intact and politically activated — a configuration the spoiler-problem literature treats as among the most destabilising, and one for which the responsibility lies substantially with the RPF's insistence on exclusion and the international mediators' acquiescence in that insistence.

The account concludes that the Arusha settlement, on its terms, was a peace agreement only the RPF could implement and only the akazu could survive — and that the failure was therefore structural rather than the product of deliberate sabotage.

10.2 The RPF-and-Kagame Account

The second account, broadly canonical in post-1994 Rwandan state historiography and supported in significant parts by the work of Des Forges (1999), Melvern (2004), and the ICTR Bagosora and Media Trial judgements, holds that the Arusha settlement was a serious good-faith negotiation that the akazu faction deliberately sabotaged through an integrated strategy across August 1993 – April 1994 while preparing the alternative of genocide. Five claims constitute this account.

First, the procedural obstruction of the BBTG installation across the eight months between signature and collapse was not a series of contingent failures but a deliberate strategy: the repeated rescheduling and cancellation of the swearing-in ceremonies, the dispute over CDR participation, the splits of the opposition parties along Hutu Power lines, and the 5 January 1994 partial swearing-in were all elements of a coordinated obstruction.

Second, the propaganda apparatus — the launch of RTLM on 8 July 1993 (twenty-seven days before signature, in clear textual violation of the Rule of Law protocol), the continuous publication of Kangura with explicit ethnic-extremist content, the December 1990 "Hutu Ten Commandments", and the cumulative incitement narrative through April 1994 — was a deliberate preparation of the population for the planned violence, not a contingent radicalisation.

Third, the militia preparation — the arming and training of the Interahamwe and Impuzamugambi, the distribution of weapons to communes, the compilation of lists, the dispersal of caches across Kigali — was the operational architecture of the planned genocide; the timing of these preparations across late 1993 and early 1994 is the principal documentary evidence (combined with the 11 January 1994 Dallaire cable's specific warning) that the akazu faction was preparing genocide while procedurally obstructing Arusha.

Fourth, the shoot-down of 6 April 1994 was the operational trigger of the prepared killing — undertaken by the akazu-aligned faction within the FAR to remove Habyarimana as the political obstacle to the prepared violence. This claim is supported by the Mutsinzi Report (2009) and broadly by the Trévidic-Poux findings (2012); it is contested by the RPF-responsibility thesis of the Bruguière indictments; the corpus does not adjudicate.

Fifth, the interim government of 8–9 April 1994 — Sindikubwabo, Kambanda, the Hutu Power coalition that took office within forty-eight hours of the shoot-down with personnel drawn from the parties' Hutu Power wings — was the institutional embodiment of the Arusha-repudiation strategy; it was prepared rapidly because the personnel and the political coalition had been organised during the obstruction period.

The account concludes that the Arusha settlement could have been implemented but was deliberately destroyed; the failure was therefore conspiratorial rather than structural.

10.3 The International-Mediator-and-Dallaire Account

The third account, represented by Jones (2001), Adelman and Suhrke (1999), Dallaire (2003), the Carlsson Report (1999), Kuperman (2001), and the broader academic literature, reads the failure as a combination of structural fragility in the protocol design and contingent diplomatic failure in the post-signature implementation. Six claims constitute this account.

First, the CDR exclusion was a structural design choice that generated rather than managed the principal spoiler — a classic textbook failure of peace-process design that the spoiler-problem literature would subsequently formalise. The exclusion was justified on rule-of-law grounds but was a political-strategic error.

Second, the speed of the transition — a 22-month timetable for the implementation of an institutional architecture more complex than the South African or Mozambican comparators — exceeded the political-institutional capacity of the parties to absorb. A longer transitional timetable with phased implementation would have given the moderate factions more time to consolidate against the Hutu Power mobilisation.

Third, the UNAMIR mandate was institutionally inadequate to the security requirements of the transition. The reduction from Dallaire's recommended 4,500 troops with robust Chapter VI to the authorised 2,548 with constrained Chapter VI reflected the post-Somalia constraints on Council action and the absence of any major-power political champion for Rwanda; the consequence was a force unable to seize the arms caches, disarm the militias, or protect the moderate political leadership.

Fourth, the post-Somalia Secretariat posture of the DPKO under Annan and Riza — risk-averse, mandate-bound, communicative rather than operational — meant that warnings such as the Dallaire cable were processed through channels that produced cautious instructions rather than mandate-expansion requests to the Council. The 11 January 1994 Riza-Annan reply is the most-cited instance of this institutional pattern.

Fifth, the Burundi shock of 21 October 1993 (the Ndadaye assassination) was a substantial external event for which the Arusha framework had no resilience; the eight months between signature and shock to collapse was insufficient time for the transitional institutions to consolidate against an unexpected destabilisation.

Sixth, the observer-state engagement — Belgian, French, American — was politically inadequate to the implementation period. Belgium, the most diplomatically engaged Western state, was politically distracted by domestic politics and by the post-Mogadishu international caution; France was structurally compromised by its concurrent military support of the FAR under Operation Noroît; the United States was institutionally absorbed in PDD-25 drafting and was unwilling to invest political capital in Rwandan implementation.

The account concludes by declining to resolve the question of whether Arusha was always doomed or was sabotageable but salvageable. Jones (2001) treats both as elements of the failure; Kuperman (2001) emphasises the structural-impossibility reading; Dallaire (2003) emphasises the contingent-failure reading. The corpus follows the third account in declining to adjudicate between the first two while documenting the third's internal heterogeneity.

11. Forward View — From Arusha's Collapse to the RPF-Inherited Transition

11.1 The Textual Inheritance

Although the Arusha process was operationally terminated on 6 April 1994 and the institutions it specified were never installed, the Arusha text survived as the legal-conceptual foundation of the post-genocide transitional arrangements. When the RPF declared the Government of National Unity on 19 July 1994 under President Pasteur Bizimungu (RPF, Hutu) and Vice-President-and-Minister of Defence Paul Kagame (RPF, Tutsi), it claimed to govern under the Arusha Accords as modified by the post-genocide reality. The modifications were substantial — the MRND and CDR were excluded as parties responsible for the genocide; the Hutu Power coalition's institutional carriers were dissolved; the moderate Hutu opposition (MDR, PSD, PL, PDC) was incorporated on terms that progressively reduced its operational autonomy — but the framework was Arusha-derived. The constitutional structure, the prefecture-and-commune administrative chain, the Council of Ministers and Transitional National Assembly forms, and the rule-of-law and refugee-return protocols all carried over.

This textual inheritance is treated in detail in RW-A-01 (genocide aftermath and RPF reconstruction), RW-C-05 (the Government of National Unity 1994–2003), and RW-D-01 (the 2003 constitution that superseded the Arusha framework as the post-transitional permanent constitutional order). The corpus reads the period 1994–2003 as the post-Arusha transitional period — an operational continuation of the Arusha framework under conditions vastly different from those it had contemplated.

11.2 The Military-Integration Inheritance

The Armed Forces protocol's integration provisions were the element most-substantially inverted by the post-genocide reality. The 60:40 FAR:RPF troop ratio became, in the post-July 1994 force, essentially a 100:0 ratio in favour of the RPF — the FAR had been militarily defeated, its officer corps had fled to Zaire (where it would form the Forces Armées Rwandaises en Exil and provide much of the leadership of the post-1994 insurgency), and the integration provisions had no constituency to enforce. The 50:50 command ratio was likewise rendered moot: the post-1994 Rwandan Patriotic Army (RPA), subsequently renamed the Rwandan Defence Force (RDF, treated in RW-I-01), was an RPF-officered force from the outset, with subsequent absorption of ex-FAR personnel proceeding on terms set by the RPF rather than by the Arusha protocol. The integration provisions thus survived as the conceptual model — a single unified armed force with cross-ethnic composition — without surviving in the protocol's specific arithmetic.

11.3 The Refugee-Return Inheritance

The Refugee Repatriation protocol of 9 June 1993 had the most substantial direct operational legacy. The post-1994 return of the old-caseload Tutsi refugees — approximately 700,000 from Uganda, Burundi, Tanzania, and Zaire, returning in the months after the RPF victory — was managed substantially under the Arusha conceptual framework: right of return without restitutionary maximalism, resettlement in vacant lands, tripartite Rwanda-UNHCR-host country coordination. The post-1996 return of the new-caseload Hutu refugees from Zaire (after the AFDL invasion of eastern Zaire and the destruction of the refugee camps) was similarly conducted under principles continuous with Arusha. Susan Thomson (2013) and the UNHCR documentation of the 1994–2003 return are the principal sources; the corpus treats the continuity as the most-durable institutional legacy of the protocol architecture.

11.4 The Rule-of-Law Inheritance

The Rule of Law protocol's principles — multi-party democracy, fundamental rights, independent judiciary, anti-exclusivist ideology — were carried over into the post-1994 constitutional order in the 5 May 1995 Fundamental Law (the interim constitutional framework combining elements of the 1991 multiparty constitution, the Arusha Accords, the 1962 independence constitution, and RPF programmatic documents). The principles would be more substantially codified in the 2003 Constitution (treated in RW-D-01), with the addition of post-genocide-specific provisions (the prohibition of "divisionism" and "genocide ideology", the abolition of ethnic mention on identity documents, the Ndi Umunyarwanda reconciliation framework). The Rule of Law protocol's anti-exclusivist-ideology provision — unenforced during the Arusha period against RTLM and Kangura — became, in inverted form, the constitutional basis for the post-1994 restriction of Hutu Power-affiliated political expression, a feature the corpus treats with care in RW-J-01 (memory and the "double genocide" thesis) and in the contested-elections documents.

11.5 Arusha in Comparative Frame

The Arusha case is among the most-studied African peace processes of the 1990s and one of the most-cited illustrations of peace-process failure in the broader scholarly literature. Comparative readings (Stedman 1997; Walter 2002; Hampson 1996; Howard 2008) identify Arusha alongside the Angolan Bicesse Accords (May 1991, collapsed in October 1992 after UNITA rejection of election results) and the Burundian Arusha process (2000–2003, ultimately partly successful) as the principal African cases for the study of spoiler problems, mandate-design failures, and the limits of negotiated settlement against committed total spoilers. The corpus position is that the Rwandan case is distinctive in three respects: the density of the protocol architecture (more elaborate than most comparators), the proximity of the principal spoilers to the regime's formal leadership (the akazu faction was inside the regime, not external to it), and the catastrophic character of the failure (genocide rather than resumed civil war).

11.6 Spiral Index

The Arusha Accords document is the bridge between the civil-war and pre-1994 documents (RW-A-03, RW-D-01, RW-H-PRES-01) and the genocide and post-genocide documents (RW-B-01, RW-A-01, RW-A-02). It is presupposed by:

RW-B-01 (the 1994 genocide), for which the Arusha collapse is the proximate political condition and for which the Hutu Power coalition that took office on 8–9 April 1994 is the institutional carrier;
RW-A-01 (genocide aftermath and RPF reconstruction), for which the Arusha framework is the legal-conceptual foundation that the post-1994 Government of National Unity inherited;
RW-A-02 (Gacaca courts), for which the Rule of Law protocol's principles supplied the constitutional framework within which the transitional-justice architecture was developed;
RW-D-01 (the Habyarimana era), for which the Arusha negotiation is the diplomatic terminus and the political environment of the regime's final two years;
RW-H-PRES-01 (Habyarimana biography), for which the Arusha process is the political context of his last two years and the proximate situation of his death;
RW-H-PRES-02 (Bizimungu biography), for which the Arusha negotiation was the principal RPF diplomatic role he carried into the post-1994 presidency;
RW-H-PRES-03 (Kagame biography), for which the Arusha process is the diplomatic counterpart of the military campaign he commanded;
RW-G-01 (education and genocide memory), for which the Arusha-period propaganda architecture (RTLM, Kangura, the "Hutu Ten Commandments") is the immediate background of the genocide-memory frameworks subsequently established;
RW-R-01 (Rwanda governance books canon), for which the Arusha process and its failure are among the principal subjects of the canonical academic literature.

The narrative continues in RW-B-01, which addresses the 1994 genocide as the operational consequence of the Arusha collapse — the eight-month gap between signature and shoot-down constituting the period in which the apparatus of the genocide was assembled while the apparatus of peace was disassembled.

Sources

Bruce D. Jones, Peacemaking in Rwanda: The Dynamics of Failure (Lynne Rienner, 2001) — the definitive academic study of the Arusha process, drawing on Tanzanian, OAU, and Western diplomatic-archival sources and on participant interviews.
Alan J. Kuperman, The Limits of Humanitarian Intervention: Genocide in Rwanda (Brookings Institution Press, 2001) — the principal scholarly counterfactual analysis, with detailed reconstruction of the UNAMIR mandate negotiation and Security Council politics.
Gérard Prunier, The Rwanda Crisis: History of a Genocide (Columbia University Press, 1995; 2nd edition with new chapter 1998) — the foundational political-historical narrative covering the 1990–1994 period.
Alison Des Forges (Human Rights Watch / FIDH), "Leave None to Tell the Story": Genocide in Rwanda (Human Rights Watch, 1999) — the principal documentary reconstruction, with extensive treatment of the 1990–1994 political mobilisation alongside the Arusha process.
Linda Melvern, Conspiracy to Murder: The Rwandan Genocide (Verso, 2004; revised 2006) — investigative reconstruction with extensive treatment of UN-Secretariat and Security Council documents released after 1994.
Linda Melvern, A People Betrayed: The Role of the West in Rwanda's Genocide (Zed Books, 2000) — the principal investigative account of the international diplomatic failure.
Filip Reyntjens, L'Afrique des Grands Lacs en crise: Rwanda, Burundi: 1988–1994 (Karthala, 1994) — Reyntjens's first-generation comparative analysis, drawing on his earlier Pouvoir et droit au Rwanda (1985) constitutional history.
Roméo Dallaire (with Brent Beardsley), Shake Hands with the Devil: The Failure of Humanity in Rwanda (Random House Canada, 2003) — the UNAMIR Force Commander's first-person account, including the 11 January 1994 "genocide fax" sequence.
Howard Adelman and Astri Suhrke (eds.), The Path of a Genocide: The Rwanda Crisis from Uganda to Zaire (Transaction Publishers, 1999) — the principal multi-author scholarly volume, with chapters by Bruce Jones, Astri Suhrke, and others on Arusha and the international response.
Brent Beardsley, contributions in Dallaire's Shake Hands with the Devil and subsequent testimony — the perspective of Dallaire's chief of staff.
Carlsson Report — Report of the Independent Inquiry into the Actions of the United Nations during the 1994 Genocide in Rwanda (S/1999/1257, 15 December 1999), Ingvar Carlsson, Han Sung-Joo, and Rufus M. Kupolati.
Arusha Peace Agreement Final Text — Peace Agreement Between the Government of the Republic of Rwanda and the Rwandese Patriotic Front (4 August 1993), with the integrated Protocols on the Rule of Law (18 August 1992), Power-Sharing I (30 October 1992), Power-Sharing II (9 January 1993), Repatriation of Refugees (9 June 1993), and Integration of the Armed Forces (3 August 1993).
UNAMIR foundational documents — UN Security Council Resolution 872 of 5 October 1993 (UNAMIR establishment); the Secretary-General's report S/26488 of 24 September 1993; the 11 January 1994 cable from Dallaire to UN DPKO; subsequent UNAMIR reporting through April 1994.
Andrew Wallis, Silent Accomplice: The Untold Story of France's Role in the Rwandan Genocide (I.B. Tauris, 2006) — on the French diplomatic-military role through the Arusha period.
Susan Thomson, Whispering Truth to Power: Everyday Resistance to Reconciliation in Postgenocide Rwanda (University of Wisconsin Press, 2013) — for analytical framing of the post-Arusha settlement's social reception.
Astri Suhrke and Bruce Jones, "Preventive Diplomacy in Rwanda: Failure to Act or Failure of Actions?" in Bruce Jentleson (ed.), Opportunities Missed, Opportunities Seized (Carnegie Corporation, 2000).
Tanzanian Ministry of Foreign Affairs archives (where consulted by Jones 2001) and Organisation of African Unity / African Union International Panel of Eminent Personalities Report on the 1994 Rwanda Genocide (OAU, July 2000).
André Guichaoua, From War to Genocide: Criminal Politics in Rwanda, 1990–1994 (University of Wisconsin Press, 2015; French original 2010) — the granular political-process reconstruction of the period.

RW-A-01: Genocide Aftermath and the Post-1994 RPF Reconstruction
RW-A-02: Gacaca Courts — Origins and Architecture
RW-A-03: Hutu Revolution, First Republic, and the Roots of Genocide (1897–1973)
RW-B-01: The 1994 Genocide Against the Tutsi
RW-D-01: Habyarimana Era and the Second Republic (1973–1994)
RW-G-01: Education and Genocide Memory
RW-H-PRES-01: Juvénal Habyarimana — A Biography
RW-H-PRES-02: Pasteur Bizimungu — A Biography
RW-H-PRES-03: Paul Kagame — A Biography
RW-R-01: Rwanda Governance Books Canon
RW-G-02: back-reference added by symmetry sweep
RW-B-02: back-reference added by symmetry sweep
RW-F-02: Rwanda-Uganda Relations 1990-2026 — From the Uganda-Trained RPF Origins to the 2019-2022 Border Closure and the Doha-Era Reset
RW-G-03: Umuganda — The Rwandan Community Service Programme
RW-D-02: The RPF Party Architecture — Inkotanyi Origins, the Cell-Sector-District Cadre System, and the Developmental-State Model
RW-D-03: Rwandan Decentralisation and the Imihigo Performance-Contract System — Districts, Sectors, Cells, Imidugudu and the Subnational Delivery Architecture
RW-G-04: Girinka — The One-Cow-Per-Family Programme
RW-D-04: Rwandan Local Elections and the Mayor-Pyramid Architecture — District Mayors, Sector Executives, Cell and Village Selection, and the Imihigo Accountability Cycle
RW-F-03: Rwanda's Cabo Delgado Deployment in Mozambique (July 2021 – 2026) — The RDF Intervention Against the ISIS-Mozambique Insurgency, the EU Funding Architecture, and the Africa Peacekeeping-Export Model
RW-D-05: Rwandan National Electoral Architecture and the RPF Dominance — Presidential and Parliamentary Cycles, the Candidate-Registration Practice, the 30 Per Cent Women's Quota, and the Contested-Record on Electoral Integrity
RW-E-03: The Rwandan Mining Sector and the DRC Conflict-Mineral Question — Tantalum, Tin, Tungsten, and Gold from the iTSCi Era Through the EU Critical Raw Materials Pact and the 2025 Sanctions Reset
RW-E-04: Rwandan Agribusiness, Specialty Coffee, and the Tea Sector Strategy — NAEB Architecture, the Specialty Shift, Tea Privatisation, Horticulture Diversification, and the 2024–2026 Export Targets
RW-J-01: The Failure to Prevent — Three Accounts of Why the 1994 Genocide Was Not Stopped, and the Contested International Response