.png)
If Proven Before A Competent Court
The ONATRA port at Mbandaka sits on the Congo River in the far west of Zaire, 2,000 kilometres by road and river from the Kivus. On 13 May 1997, soldiers of the AFDL and the Rwandan army caught up with refugees who had covered that entire distance on foot, by barge and by train, and opened fire on them at the port. They fired for five to ten minutes. The UN Mapping Report cites sources putting the dead at at least 200. Many of the bodies went into the river. At Wendji nearby the same day, at least 140 more were allegedly killed.
That was the end of it. The beginning was seven months earlier and a continent's width to the east.
The camps were also a barracks
After the Rwandan genocide of 1994, UNHCR counted some 1.2 million Rwandan refugees in Zaire: roughly 850,000 around Goma, in camps called Kibumba, Katale, Kahindo, Mugunga and Lac Vert, and 332,000 more around Bukavu.
The camps also contained the people who had committed the genocide. That is not a later Rwandan talking point. It is in UNHCR's own record.
In late August 1994, High Commissioner Sadako Ogata wrote to the UN Secretary-General asking him to "totally disarm the ex-FAR troops, collect all arms and military equipment" and to "isolate and neutralize civilian leaders." By then, in UNHCR's own words, ex-FAR officers controlled the camps. Nothing of the kind was done.
Howard Adelman, in the Brookings volume Refugee Manipulation, cites a UN count of 16,000 ex-FAR and 35,000 to 50,000 militia, and quotes UNHCR officials: "We are in a state of virtual war in the camps."
From Kibumba to the river
On the night of 25 to 26 October 1996, AFDL and Rwandan army units attacked Kibumba, 25 kilometres north of Goma. The hospital was destroyed and roughly 194,000 refugees fled west towards Mugunga. Burial records logged 1,919 bodies interred by one church body over the next two months, and 2,087 buried by Kibumba residents in November alone.
Katale fell on 26 October; of 970 later buried there, "many bodies were found in the public latrines." Kahindo and Katale were attacked again on 31 October. Roberto Garretón, the UN Special Rapporteur, later counted 100 to 200 dead at Kahindo and 143 at Katale.
South Kivu was identical. On 26 October, between 200 and 600 people were killed on the roads between Nyantende and Walungu, among them women, children and the elderly. On 2 November, hundreds died at Kashusha and INERA. At Chimanga camp, on or about 22 November, between 500 and 800 refugees were killed. The soldiers had first promised they would slaughter a cow and hand out the meat.
Mugunga was shelled with heavy weapons for six hours on 14 November. The next day a journalist counted 40 dead along the Sake road, "including women, children and two babies," and Red Cross volunteers buried 166 more. At Lac Vert, Hutu men were bound and thrown alive into the water.
Then the camp population split. Several hundred thousand walked east, back into Rwanda. The rest walked west, and were followed.
Maniema, March 1997. AFDL and Rwandan units entered the Tingi-Tingi site outside Lubutu on 1 March and killed several hundred of its remaining occupants, including the sick in the dispensary and unaccompanied children. At the Lubilinga river bridge the same day they shot several hundred more who were queuing to cross.
Then something colder. Between 14 March and 2 April 1997, at Tingi-Tingi and Amisi, AFDL/APR soldiers "deliberately blocked all humanitarian, health and medical aid." At Tingi-Tingi at least 216 refugees died. That is not a massacre. It is a decision taken and held for twenty days.
Kisangani, April 1997. At the Kasese camps, villagers attacked with machetes and arrows on 21 April, "visibly encouraged by AFDL/APR soldiers"; the next day AFDL/APR soldiers killed at least 200 there, buried them, then exhumed and burned them.
Then Équateur, and the river, and Mbandaka.
Three hundred thousand missing is not three hundred thousand dead
You will read that 200,000 to 300,000 refugees were killed in Zaire. The upper figure has no institutional source as a death toll, and its provenance is worth tracing, because it began as a missing-persons tally, not a body count.
The UN's summary of Sadako Ogata's New York press briefing of 28 April 1997 reads:
"Beside the 80,000 refugees that had recently disappeared south of Kisangani, another estimated 170,000 Rwandan refugees were still missing from other camps. When added to the 40,000 Burundian refugees that were somewhere in Zaire, that came to a total of 300,000 refugees missing in the country."
Missing. Not dead. Ogata was describing people whose whereabouts UNHCR could not establish, in a country where hundreds of thousands were moving through forest without registration. Her components sum to 290,000. The UN's own summary rounded up.
Human Rights Watch, that October, cited UNHCR's 213,000 Rwandans "unaccounted for," and defined the term in the same sentence: "either dead in the period of violence or hidden in the forests or among the people of Congo." Le Pape and Bradol of MSF's research arm CRASH note that the 200,000 now circulating as a count of the vanished was originally an NGO estimate of victims, and that estimates of the number in flight run from 400,000 to 500,000.
The UN's own number for the dead, when it finally produced one, is narrower: "probably several tens of thousands, all nationalities combined."
The investigation that was sent home twice
Two UN bodies tried to establish what happened while it was still happening, and the popular account collapses them into one.
The first was a joint mission of three human rights mandates, Garretón's among them, reporting as document A/51/942 on 2 July 1997. It listed 134 alleged massacres and found that "according to the information received, several thousand people were killed, including large numbers of women and children." On the central question it wrote only that "the question of whether a genocide was planned and executed should be asked." It had returned to Geneva without reaching the area it was sent to investigate.
The second was the Secretary-General's Investigative Team, created in July 1997 and headed not by Garretón but by Chief Justice Atsu-Koffi Amega of Togo. Kinshasa objected to its composition, organised hostile demonstrations against it, and in April 1998 detained one of its investigators and seized his documents. The team withdrew for good on 17 April. Its report concluded that the massacres and the denial of aid "were systematic practices involving murder and extermination, which constitute crimes against humanity," and that "one possible interpretation" was "that a decision was taken to eliminate this part of the Hutu ethnic group as such. If proved, this would constitute genocide."
Damning became inculpatory
The mapping exercise that finally did the counting had its terms of reference approved in May 2007, under High Commissioner Louise Arbour. Around twenty human rights officers deployed between October 2008 and May 2009, to cover a decade across a country the size of western Europe. They interviewed more than 1,280 witnesses and documented 617 violent incidents, of which 104 were attacks on Hutu refugees and Hutu civilians.
The report was finished in June 2009 and then sat for over a year.
On 3 August 2010, Rwandan Foreign Minister Louise Mushikiwabo wrote to Ban Ki-moon that "attempts to take action on this report, either through its release or leaks to the media, will force us to withdraw from Rwanda's various commitments to the United Nations, especially in the area of peacekeeping." Rwanda had thousands of troops in UN missions, including in Darfur.
Three weeks later, on 26 and 27 August, Le Monde published the leaked draft. The leak came after the threat, not before it. The draft had gone to governments in the region, and nobody has established who leaked it.
Navi Pillay postponed publication by a month on 2 September, before Ban flew to Kigali: "Following requests, we have decided to give concerned States a further month to comment on the draft." She offered to publish those comments alongside the report, and on 1 October 2010 she did.
The draft had said the attacks "reveal a number of damning elements that, if they were proven before a competent court, could be classified as crimes of genocide." The published text says "inculpatory elements" that "could be characterized as crimes of genocide."
Al Jazeera reported three changes on publication day: that substitution, an expanded countervailing factors section, and the removal of a draft passage dismissing mitigating arguments. Colum Lynch in Foreign Policy printed before and after pairs showing inserted conditionals and the adverbs "allegedly" and "apparently." Human Rights Watch, the same day, said flatly that there was "no substantive difference."
Both are defensible. The vocabulary was softened and the counter-case was enlarged. The genocide sentence survived, and it is in the published report.
The countervailing factors
Here is the part almost nobody quotes, and the piece is worthless without it. Immediately after the genocide sentence, the report lists reasons a court might find the other way.
Intent to destroy the group "as such" must be proved, and "it does not suffice to prove that members of the group were targeted because they belonged to the group." The killings might instead constitute "collective retribution against Hutu civilians in Zaire suspected of involvement with the ex-FAR/Interahamwe." And "whilst in general the killings did not spare women and children," in some places, "particularly at the beginning of the first war in 1996, Hutu women and children were apparently separated from the men, and allegedly only the men were subsequently killed."
And the one that carries the most weight: "Facts which tend to show that the RPA/AFDL spared the lives, and in fact facilitated the return to Rwanda of very large numbers of Hutu militate against proving a clear intent to destroy the group."
That is not a rhetorical hedge. UNHCR recorded 380,000 returnees crossing from Goma into Gisenyi in the first movement after Mugunga fell, and perhaps 500,000 to 600,000 with later crossings.
Rwanda's formal comments of 30 September 2010 build on exactly that, and they build well. The camps housed an armed force preparing "a major assault on Rwanda." The exodus itself was "by all standards a political-military strategy," with civilians brought along as human shields. And seven months of field work cannot carry a genocide finding across a decade and a country ninety times Rwanda's size on what Kigali called the "absolute lowest evidentiary standard."
The report concedes the human shield point in its own executive summary: the refugees, "who it appears were often rounded up and used by the ex-FAR/Interahamwe as human shields during their flight."
No one is known to have been tried
The International Criminal Court cannot touch any of this. Its jurisdiction over the DRC begins on 1 July 2002, and the DRC's own 2004 referral covers the situation "since 1 July 2002." The crimes predate the court's reach by five years.
The remedy most pressed since has been a mixed chamber inside the Congolese system, judges national and international. The bill died in the Senate in 2011, and the National Assembly voted it inadmissible on 8 May 2014, members objecting to provisions allowing prosecution of MPs and soldiers. Nothing has replaced it.
Spain came closest. On 6 February 2008 Judge Fernando Andreu Merelles of the Audiencia Nacional indicted 40 senior Rwandan officers, reaching the Congo killings as well as events in Rwanda, on the hook of nine Spanish nationals killed or disappeared. General Emmanuel Karenzi Karake was arrested at Heathrow on the resulting warrant in June 2015, and the extradition was dismissed that August on a jurisdictional argument. In 2015 the case was shelved under a 2014 law that narrowed Spain's universal jurisdiction. In June 2017 the National Court partly reopened it over the murders of Spanish nationals.
No one is known to have been tried anywhere for the killing of Rwandan Hutu refugees in Zaire. A 2025 survey by TRIAL International and UpRights counted more than 130 Congolese judgments on international crimes since 2004.
It was never a failure of warning. Ogata gave her missing figure in April 1997, the joint mission listed 134 massacres in July, Human Rights Watch published in October, and the Secretary-General's own team reached crimes against humanity by June 1998. Accountability failed. Knowledge did not.
The UN's last word on the question it raised is paragraph 33 of its own executive summary, and it reads as an instruction nobody has carried out:
"In light of the competing considerations reviewed above, a full judicial investigation into the events that occurred in Zaire in 1996 to 1997 will be necessary, in order to permit a competent court to decide on the matter."
Sources consulted
OHCHR, DRC Mapping Report, 1 October 2010, Executive Summary paras 27 to 33, and factsheets FS-2 and FS-6
Government of Rwanda, comments on the draft Mapping Report, 30 September 2010
UN doc S/1998/581, Secretary-General's Investigative Team, 29 June 1998
UN doc A/51/942, joint mission on massacres in eastern Zaire, 2 July 1997
UN press briefing by UNHCR, New York, 28 April 1997
UNHCR, The State of the World's Refugees 2000, chapter 10
Human Rights Watch, What Kabila is Hiding, 1997, and its Mapping Report Q&A, 2010
Howard Adelman in Refugee Manipulation, Brookings, 2003
Le Pape and Bradol, MSF-CRASH, 30 March 2021
Al Jazeera and Foreign Policy, 1 and 2 October 2010; AP, 28 August 2010
ICTJ, 2015; UpRights and TRIAL International, 2025; ICC DRC situation page
Next in Cold Files: Biafra, 1967 to 1970. A blockade, a famine broadcast into living rooms, and a death toll that ranges across an order of magnitude.
Cold Files runs twice a week alongside the morning brief. For both, join us on Telegram at t.me/TheLeaflet.
Comments