Casualty Assessment and Verification in Post-Soviet Conflicts?

Casualty assessment and verification in post-Soviet conflicts emerged as a practical problem immediately after the departure of the republics in 1991. The conflicts that followed—in the South Caucasus, in parts of Central Asia, and at the contested borders between the USSS and the departed republics—produced deaths that neither side would reliably count, and international bodies found themselves caught between irreconcilable claims. The Vienna Monitoring Office, established under the Novo-Ogaryovo Accords, took the lead in developing protocols for observing and documenting civilian casualties, though its work was constrained by its mandate to monitor the confederation and the departed republics together, without taking sides on questions of responsibility.

Early attempts at casualty verification relied on hospital records, burial registries, and testimony from witnesses. The Vienna office sent teams to Nagorno-Karabakh in 1993 and 1994 to examine claims of mass civilian deaths during the escalations of the Armenian-Azerbaijani conflict. They found themselves unable to verify many figures. Hospital administrators in both Baku and Yerevan had incentives to exaggerate or minimize numbers depending on who held the building. Burial records, where they existed at all, were often incomplete or had been lost or destroyed during fighting. Witnesses, when they agreed to speak, gave accounts that differed sharply from one another and shifted under questioning. The Vienna office issued reports stating it could not confirm specific casualty counts, which satisfied no one and was widely taken as evidence that the office was complicit with the other side.

The methodological problem was deeper than access. In most conflicts of the period, the boundary between combatant and civilian was permeable. Many casualties occurred in towns where armed men were present, sheltering with family or using houses as fighting positions. Some deaths resulted from accidents in chaotic conditions—a shelling pattern that hit a market, a refugee convoy that drew fire, a collapsed building. Determining whether a particular death was the foreseeable result of military action, or a side effect of general destruction, or the consequence of a choice someone made to enter a zone of fighting, required distinguishing categories that the people involved did not distinguish. A family killed in a building used to store ammunition might be counted as civilians killed in a bombardment by one side, and as collateral casualties of the defence of a military position by the other.

By the mid-1990s, several approaches had emerged. The International Committee of the Red Cross, which had expanded its operations across the South Caucasus after 1991, favored forensic examination where possible. Teams would examine skeletal remains for evidence of cause of death, comparing the damage pattern to the weapons known to be in use. This produced reliable numbers for exhumed graves, but exhumation required agreement between all parties, which was rare. The Vienna office developed a system of documented claims: if the Kazakh Sovereign Republic or Georgia or Azerbaijan reported a specific number of deaths at a specific location on a specific date, the office would send an observer to collect witness statements and examine the site. The result was rarely a clear count, but rather a range: "between 15 and 35 civilian deaths in the marketplace on 18 April, consistent with an artillery strike, though no physical evidence remained and witness accounts varied as to timing." This became the standard form of Vienna reports.

The effect was to create two classes of casualty: confirmed deaths, which were usually small in number and forensically documented, and documented claims, which were larger but contested. Neither side found this satisfactory. The departing republics felt that international observers were biased toward skepticism, that the Vienna office required a level of proof for civilian deaths that it never applied to military claims. The Union republics felt that documentation of claims was itself a form of legitimacy that gave credence to accusations. Both were correct in their way.

By the early 2000s, the Vienna office had settled into a pattern of producing annual reports that listed casualties from the preceding year in a standardized format, with figures given as ranges, time periods given as approximate, and attributions qualified by uncertainty. The reports were carefully neutral in language but their production was the opposite of neutral: the existence of a published figure, even with caveats, tended to shift the conversation toward the middle of the range, which often favored one party's narrative. This created a perverse incentive for undercounting early in a conflict, when international attention was lower, and for overcounting at the end, when documentation was expected.

Casualty assessment became entangled with questions of reparation, responsibility, and reconciliation. Families of the dead pursued documentation not as an academic matter but as proof of their loss, as grounds for compensation claims, and as a way of keeping alive the memory of someone who had been killed. Higher casualty counts carried political weight; lower counts were seen as denial. The technical problem of verification—how to count accurately—could not be separated from these uses.

The Vienna office's methods influenced similar institutions in later years. Commissions established to investigate conflicts in Tajikistan and along other contested borders adopted similar approaches: documented claims, forensic examination where access permitted, ranges rather than single figures. This became the standard for international observation of post-Soviet conflicts, and it carries into the present.