Khojaly massacre §

Background

By February 1992 the war over Nagorno-Karabakh had escalated from the late-Soviet pogrom period into open inter-state conflict. The autonomous oblast had been formally abolished by the Azerbaijani parliament in November 1991 (the abolition of NKAO); Karabakh Armenians had voted overwhelmingly for independence on 10 December 1991 (the Karabakh referendum); and the Stepanakert Armenian population had spent the winter under continuous artillery bombardment from the Azerbaijani-held heights of Shusha and from Khojaly itself (Human Rights Watch, 1992, pp. 14–17). By Memorial's count, several hundred Stepanakert civilians had been killed in indiscriminate Grad MLRS and artillery fire from these positions during the winter siege (Memorial Human Rights Center, 1992b). The operational rationale that the Armenian command itself articulated for the assault on Khojaly — to silence the bombardment of Stepanakert and to open the airport for humanitarian flights — does not absolve the conduct of the attack, but it is the operational context that the most-cited international accounts sometimes elide editorial.

Khojaly was the only town in the former NKAO with an ethnic Azerbaijani majority and the location of the autonomous oblast's only airport (Thomas de Waal, 2003, pp. 169–172). It sat astride the road from Stepanakert to the Armenian-populated district of Askeran. Through January and February 1992, Armenian forces had been tightening a siege of Khojaly; the airport had ceased to function in late January, the town's electricity, water and food were cut off, and a 3 February attempt by an Mi-26 helicopter to evacuate civilians ended when the aircraft was shot down near the town with all aboard killed (Thomas de Waal, 2003, p. 170). The defending Azerbaijani garrison consisted of the local OMON detachment, militia volunteers from the Popular Front (around 160 fighters), and several armoured vehicles (Thomas de Waal, 2003, pp. 170–171). The civilian population at the time of the attack was probably between 2,500 and 3,000, substantially below the pre-conflict figure of 6,300 (Memorial Human Rights Center, 1992a, Demographic prelude). International evacuation talks brokered through the OSCE and Russian channels in early February 1992 had failed without producing safe passage for the population (Thomas de Waal, 2003, pp. 170–172).

The Armenian attacking force included units of the nascent NKR Defence Army and personnel of the Russian 366th Motor Rifle Regiment, which had been formally a CIS unit since the dissolution of the USSR but which had de facto come under Armenian operational control during the late Soviet collapse, a fact later documented by Human Rights Watch and by Russian veterans' testimony (Human Rights Watch, 1992, pp. 19–22). sourced opinion The 366th was garrisoned in Khankendi (Stepanakert under its Azerbaijani name) and had been the principal Soviet military presence in the autonomous oblast since the 1980s; its command status between December 1991 and March 1992 was nominally CIS, ambiguously Russian, and operationally enmeshed with NKR Armenian forces (Human Rights Watch, 1992, Ch. 2). sourced opinion The Russian Defence Ministry would formally withdraw the regiment in March 1992, partly in response to international criticism following the Khojaly assault (Thomas de Waal, 2003, p. 175).

The Armenian operational chain at the time of the assault is editorially load-bearing and was for years under-emphasised in Armenian public discussion. Robert Kocharyan was head of the NKR State Defence Committee; Serzh Sargsyan was the operational commander of NKR forces in Karabakh; de Waal establishes both as in command at the time of the Khojaly assault (Thomas de Waal, 2003, pp. 174, 184–186). Both men later became Presidents of Armenia (Kocharyan 1998–2008; Sargsyan 2008–2018, ousted in the Velvet Revolution).

The event

The assault began with artillery preparation on the evening of 25 February 1992 and ground assault overnight. By dawn on the 26th, the town was under Armenian control. The Armenian command had announced via radio that a humanitarian corridor would be opened to allow civilians to evacuate eastward toward Aghdam (Memorial Human Rights Center, 1992a, Reconstruction §3).

What happened next, and how to characterise it, is the central interpretive question of the war (Thomas de Waal, 2003, pp. 172–178). contested The civilian column, mixed with retreating fighters, attempted to cross the snow-covered fields between Khojaly and the Azerbaijani-held positions east of the river. Some of the column was shot at close range as they crossed open ground near the village of Nakhichevanik (Memorial Human Rights Center, 1992a, Findings §4–§5). Others, particularly the elderly and children, died of cold during the multi-day flight. Some were killed in confused close-quarters combat between the column and Armenian units that may or may not have known of the corridor.

The contemporaneous investigation by Memorial Human Rights Center (Moscow), conducted in March 1992 with access to the bodies and to survivors, documented at least 161 civilian dead and concluded that the killings were not the spontaneous product of close-quarters combat alone, that fleeing civilians were deliberately killed, and that the corridor was inadequately marked or partially blocked (Memorial Human Rights Center, 1992a, Tally §6; Memorial Human Rights Center, 1992a, Findings §4–§5). The Memorial report attributed the killings to Armenian forces, with the qualification that the involvement of the Russian 366th regiment was probable but unverified. Human Rights Watch reached similar conclusions and characterised the event as "the largest massacre to date in the conflict" (Human Rights Watch, 1992, p. 5). The site of the densest civilian killing was the open field outside the village of Nakhichevanik, north of the announced corridor; Memorial's investigators interviewed survivors who said they did not know where the corridor was, and HRW's interviews with Armenian fighters produced parallel statements that the corridor's existence was not communicated through the chain of command (Memorial Human Rights Center, 1992a, Findings §4).

Chingiz Mustafayev, the Azerbaijani-Russian war journalist, filmed two passes through the kill zone north of Khojaly, on 28 February and 2 March 1992 (Chingiz Mustafayev (footage) & compilers, 1992). The discrepancies between his two films — particularly in the position and condition of bodies, and in evidence of post-mortem mutilation visible on the second pass that was not visible on the first — became the evidentiary core of later Armenian-side denial of mutilations. The Mustafayev footage globalised Khojaly: it ran on Russian state television, in Western coverage, and in Azerbaijani state media in the weeks immediately following the assault. Mustafayev himself was killed in fighting near Aghdam in June 1992, four months after his footage went global; he was posthumously named National Hero of Azerbaijan.

The Azerbaijani government's official figure, originally announced by President Ayaz Mutallibov himself on 13 March 1992, is 613 dead, including 106 women, 63 children and 70 elderly (Thomas de Waal, 2003, pp. 175–176). Modern Azerbaijani sources extend the toll. The factual basis for the official figure, which has remained unchanged since 1992, has not been transparently documented, and independent observers have generally adopted the Memorial range of 200–500, with de Waal settling at approximately 485 from hospital records and witness interviews (Thomas de Waal, 2003, pp. 172–178). sourced opinion The toll itself remains contested between state, NGO and academic sources (Thomas de Waal, 2003, p. 175). contested

Approximately 30 Azerbaijani combatants were also killed in the defence and the breakout, separately documented by Memorial and HRW alongside the civilian toll (Memorial Human Rights Center, 1992a, Tally §6); this combatant figure does not enter the disputed range above and does not affect the international-humanitarian-law characterisation of the event, but its absence from the most-circulated tallies is itself a fact about how the atrocity has been presented editorial.

Numbers

Source Figure Method Memorial (1992) ≥ 161 Field investigation, body and survivor access, March 1992 HRW (1992) "several hundreds" Field investigation; named-unit responsibility de Waal (2003) ~485 Hospital records + witness interviews Azerbaijani government (announced 13 March 1992) 613 Original presidential statement; basis not transparently disclosed Azerbaijani combatant deaths (separate) ~30 Memorial + HRW; rarely included in commemorative tallies

Approximately 1,275 people were taken hostage during and after the assault; ~150 remained unaccounted for on contemporaneous reckonings, with most of the rest exchanged through the Russian-mediated Kazimirov channel over the following year (Memorial Human Rights Center, 1992a, Tally §7).

The corridor evidence

Of the central evidentiary controversies of the First Karabakh War, the Khojaly humanitarian corridor is the most documented and the most disputed. Three contemporaneous evidentiary streams converge on the same compound finding editorial:

• Memorial's March 1992 reconstruction mapped the announced corridor against the location of the bodies and concluded that the corridor was either inadequately marked, partially blocked, or in places unilaterally closed by Armenian fighters who had not been informed it existed (Memorial Human Rights Center, 1992a, Reconstruction §3, Findings §4–§5). The corridor on the Memorial map ran from the eastern edge of Khojaly toward Aghdam through fields north of Nakhichevanik; the densest cluster of bodies was approximately 200 metres south of the corridor's announced line.
• Mutallibov's April 1992 interview with Dana Mazalová in Nezavisimaya Gazeta explicitly stated that the corridor had existed and that the bodies near it "were shot with deliberate cruelty" (Thomas de Waal, 2003, pp. 175–177 (quoting Nezavisimaya Gazeta, 2 April 1992)). This is the most consequential single Azerbaijani-side primary source for the corridor's existence.
• HRW's interviews with Armenian fighters produced statements consistent with the gap between political-level corridor announcement and operational-level corridor implementation (Human Rights Watch, 1992, Ch. 2).

The "Aroshyan boys" attribution from Markar Melkonian's My Brother's Road (2005) — Monte's posthumous biographical account naming a specific Armenian unit responsible for second-day killings beyond Nakhichevanik — is the closest the war's Armenian-language literature has come to internal attribution (Melkonian, 2005). sourced opinion The book's source-quality is mixed (memoir-biography, partly second-hand) but the candour about Armenian-side conduct is unusual and the attribution is consistent with both the Memorial map and HRW interviews.

Aftermath

The political consequences were immediate and decisive. President Mutallibov, in a remarkable interview with the Czech journalist Dana Mazalová published in Nezavisimaya Gazeta on 2 April 1992, stated that the corridor had existed and that the bodies near the corridor "showed that the people were shot with deliberate cruelty" (Thomas de Waal, 2003, pp. 175–177 (quoting Nezavisimaya Gazeta, 2 April 1992)), and in the same interview implied that the Azerbaijani Popular Front, then in opposition, had let the killings happen in order to bring his government down (Thomas de Waal, 2003, pp. 176–177). sourced opinion The interview cost Mutallibov his presidency: he resigned on 6 March (in a first round of pressure) and again on 14 May (after a brief restoration), opening the way for the Popular Front under Abulfaz Elchibey (Audrey L. Altstadt, 2017, Ch. 7).

The Russian 366th Motor Rifle Regiment was formally withdrawn by the Russian Defence Ministry in March 1992. Russian sources name Colonel Yuri Zarvigorov as having had operational coordination with NKR forces during the assault; the Russian state has never published a formal investigation of the regiment's role and the Russian-archival material on Khojaly remains substantially closed (Thomas de Waal, 2003, pp. 173–175). sourced opinion

Khojaly entered Azerbaijani public memory as the country's foundational war atrocity and from the late 1990s under Heydar Aliyev and afterward under Ilham Aliyev became the centrepiece of an internationalised "Justice for Khojaly" diplomatic campaign run from Baku through the diaspora and the Heydar Aliyev Foundation under Mehriban Aliyeva (Thomas de Waal, 2003, Postscript (2013 ed.)). The campaign's unifying graphic, a child's-hand silhouette with one finger missing, designed by Berlin-based Azerbaijani photographer Sanan Aleskerov in 2009, became the visual register of the diplomatic effort (Thomas de Waal, 2003, Postscript).

The state-level recognition campaign expanded substantially in the 2010s. Mexico's federal congress passed a recognition resolution in 2011, Pakistan's Senate in 2012 with explicit "genocide" framing, Turkey's parliament in 2010, the Czech Senate in 2013, Romania's parliament in 2015, and several US state legislatures (Texas, Massachusetts, New Mexico among others) over the same period (Thomas de Waal, 2003, Postscript). The European Parliament's 2010 resolution on the South Caucasus referred to the Karabakh war's atrocities including Khojaly under the framing of "massive ethnic cleansing" without using the "genocide" formulation (Parliament, 2010). The "genocide" characterisation runs against the assessment of the international human-rights community, which has held to the war-crime / crime-against-humanity classification: the killings, while clearly rising to grave-violation status under the Geneva Conventions, do not meet the 1948 Genocide Convention's group-destruction-with-intent test that requires proof of intent to destroy in whole or in part a national, ethnic, racial, or religious group as such (Thomas de Waal, 2003, Postscript). contested

Azerbaijan filed a contentious case against Armenia before the International Court of Justice in September 2021 under the 1965 Convention on the Elimination of All Forms of Racial Discrimination; the application references Khojaly among the war's atrocity events but pleads a pattern of racial discrimination across 1988-2020 rather than seeking a standalone genocide characterisation (International Court of Justice, 2021a). sourced opinion The ICJ's interim provisional-measures orders of December 2021, February 2023, July 2023 and November 2023 addressed humanitarian-corridor and protection-of-Armenian-residents issues but did not engage the Khojaly events specifically (International Court of Justice, 2021b). The case remains pending on the merits at the time of writing.

Historiography

The academic literature on Khojaly converges on a compound finding sourced opinion: fleeing civilians were deliberately killed at close range along an announced humanitarian corridor that was inadequately marked, partially blocked, or unknown to forward Armenian units; the toll lies between the Memorial floor of 161 and the de Waal reconstruction of approximately 485; the Russian 366th Motor Rifle Regiment's operational role was material to the assault though its specific conduct in the corridor killings has not been transparently documented; and the international-legal characterisation is war crime and crime against humanity rather than genocide under the 1948 Convention (Thomas de Waal, 2003, pp. 169-183, Postscript).

de Waal's Black Garden (2003) reconstruction combined hospital records, named-victim lists from both Armenian and Azerbaijani sources, and approximately 120 interviews with combatants and survivors on both sides. de Waal settled at roughly 485 confirmed dead and described the corridor in operational terms as announced at the political level, partially marked on the ground, and at the Nakhichevanik crossing closed unilaterally by Armenian fighters who had not received the corridor order (Thomas de Waal, 2003, pp. 169-183). sourced opinion Cornell's Small Nations and Great Powers (2001) treats the 366th's involvement as militarily decisive — without Russian armour and artillery preparation, the Khojaly assault would not have been operationally possible — but locates responsibility for the corridor killings on the Armenian side rather than as Russian-directed (Svante E. Cornell, 2001, Ch. 5). sourced opinion Cheterian's War and Peace in the Caucasus (2008) frames the corridor failure as a deliberate political choice on the Armenian side rather than a logistical breakdown, citing the timing of the corridor announcement, the unit positions immediately before the assault, and the absence of any internal-investigation evidence afterwards (Cheterian, 2008). sourced opinion Broers's Armenia and Azerbaijan: Anatomy of a Rivalry (2019) integrates these readings into a longer-frame analysis of the war's atrocity register and treats Khojaly as the single best-documented and least-investigable event of the conflict's atrocity record (Broers, 2019). sourced opinion

The Azerbaijani state historiography, consolidated under the Heydar Aliyev presidency from 1993, expanded the toll beyond the Memorial range without publishing the primary evidence base, characterised the killings as "genocide" rather than war crime, and elided the Russian 366th's role in favour of attribution to Armenia and the Karabakh Armenian leadership alone (Thomas de Waal, 2003, Postscript). (sourced opinion, disputed) The methodology has been criticised by Western academic reviewers (de Waal, Cornell, Broers) for the absence of body-count primary documentation and the absence of forensic-anthropology investigation in the post-1992 period, when Azerbaijani access to the site would in principle have permitted comprehensive documentation (Thomas de Waal, 2003, Postscript). sourced opinion The 2020 Azerbaijani recapture of the Khojaly area in the Second Karabakh War reopened the possibility of on-site forensic investigation by independent international teams; as of this writing none has been conducted or announced (Broers, 2019). editorial

The Russian-archival picture remains substantially closed. The 366th Motor Rifle Regiment's operational logs for February 1992 have not been published by the Russian Ministry of Defence; the named regimental commander at the time, Colonel Yuri Zarvigorov, has not given on-the-record testimony; Russian veterans' testimony, fragmentary and partly contradictory, has been collected in journalistic accounts and in human-rights documentation but does not constitute archival reconstruction (Thomas de Waal, 2003, pp. 174-178). Until the Russian archive opens, an event of no near-term political possibility, the operational reconstruction of the 366th's specific role in the assault and the killings rests on Memorial's 1992 interview record, the Mustafayev footage, and the published combatant testimony on both sides (Memorial Human Rights Center, 1992a). editorial

What is not contested in the academic literature: the deaths of fleeing civilians at close range on the night of 25-26 February and into the following days; the existence of an announced corridor that failed in execution; the responsibility of Armenian forces (with 366th support) for both the assault and the killings; the rejection of the "spontaneous close-quarters combat" framing as adequate to explain the casualty pattern. What remains contested in the academic literature: the precise toll (Memorial 161 floor, de Waal ~485, Azerbaijani state 613); the chain of command for the corridor failure (political-level decision, operational-level breakdown, or local-unit initiative); the extent of post-mortem mutilations visible on the second Mustafayev pass but not the first; and the international-legal characterisation of the event (Thomas de Waal, 2003).

International law

The killings of 25-26 February 1992 occurred during an international armed conflict between Armenia and Azerbaijan over the territory of Nagorno-Karabakh, with the Soviet Union dissolved and the successor states formally independent and recognised since December 1991. The Fourth Geneva Convention, its Additional Protocols I and II of 1977, and the customary international law of armed conflict therefore applied at the time (High Contracting Parties to the Geneva Conventions, 1949). The deliberate killing of civilians hors de combat, the killing of those fleeing along an announced corridor, and the close-range targeting of unarmed civilians are war crimes under the framework operative at the time, and were subsequently codified in the Rome Statute (Article 8) when the ICC came into force in 2002 (High Contracting Parties to the Geneva Conventions, 1949). Neither Armenia nor Azerbaijan is a party to the Rome Statute; the ICC therefore has no jurisdiction over Khojaly absent UN Security Council referral, an outcome of no near-term political possibility (Human Rights Watch, 1992). editorial

The international human-rights and NGO communities — Memorial, Human Rights Watch, Amnesty International — have characterised the events as a war crime and as conduct rising to crime-against-humanity status, but have not endorsed the "genocide" framing pursued by the Azerbaijani state recognition campaign (Memorial Human Rights Center, 1992a). sourced opinion The 1948 Genocide Convention's test requires proof of "intent to destroy, in whole or in part, a national, ethnical, racial or religious group as such"; the academic and NGO assessment has been that the operational intent at Khojaly, while clearly criminal, was to capture the strategic position and silence the Stepanakert bombardment, not to destroy the Azerbaijani population as such (Thomas de Waal, 2003). sourced opinion The distinction matters legally and politically: war crime / crime against humanity invokes individual command-level accountability with no statute of limitations; genocide invokes additional state-level responsibility and broader symbolic register. The Azerbaijani state has pursued the genocide framing for the symbolic register; the international academic and NGO consensus has resisted on the intent test.

Memory and politics

Khojaly is commemorated annually on 26 February in Azerbaijan and is one of the two atrocities most commonly cited in Azerbaijani-genocide-recognition diplomacy. In Armenian discourse the tendency has run between three poles: (i) recognising civilian deaths but disputing organisation and intent; (ii) blaming the deaths on the chaotic exit and Mutallibov's calculated abandonment; and (iii) attributing the bulk of the killings to the Russian 366th regiment in order to limit Armenian responsibility (Thomas de Waal, 2003, pp. 172–178). contested None of these framings has been embraced by the international NGO and academic mainstream, which has held to the Memorial assessment.

The Armenian state-level position evolved across three decades. Through the 1990s the Armenian government and the Karabakh Armenian leadership emphasised the corridor's existence, the Mutallibov-Popular Front political dynamic, and the 366th's involvement, in a register that broadly denied Armenian operational responsibility (Thomas de Waal, 2003, pp. 178-183). The Kocharyan and Sargsyan presidencies (1998-2008; 2008-2018) extended that framing, with both men personally veterans of the Khojaly operation and therefore both interested parties and witnesses (Thomas de Waal, 2003). editorial Prime Minister Nikol Pashinyan, elected in the 2018 Velvet Revolution without combat-veteran credentials and from outside the wartime political class, became the first Armenian leader to publicly acknowledge the deaths of Khojaly civilians as a tragedy requiring Armenian reflection. In a February 2019 AP interview he characterised the killings as a tragedy and declined to endorse the framings used by his predecessors, while continuing to reject the "genocide" formulation that the international academic community also rejects (Avet Demourian (AP & Yerevan), 2019, AP interview, Yerevan, 22 February 2019). sourced opinion In the post-2020 peace-treaty negotiations between Armenia and Azerbaijan, mutual atrocity recognition has been raised on both sides as a precondition for normalised relations; Khojaly and Maraga are the two events most often paired in this diplomatic framing (Broers, 2019). editorial

The asymmetry between Khojaly's international visibility and the comparable Armenian-side documentation of Maraga is structural: Maraga had earlier and more thorough field documentation through Christian Solidarity International but no Aliyev-state-equivalent diplomatic mobilisation behind it; the absence of a comparable Armenian state campaign is one editorial explanation for the visibility gap, with scale of the toll being another editorial.

The Khojaly–Sumgait–September Days–Armenian Genocide memory complex now structures both societies' understanding of the conflict. The full multi-perspective treatment, with sourced critique on each viewpoint, is at the Khojaly dispute in this atlas. editorial