352,000 deaths in four years. Mediazona and Meduza’s new estimate of Russian losses in Ukraine

Our estimation methodology is based on data from the Probate Registry and the recorded names count we are maintaining together with BBC News Russian and a group of volunteers. We’ve described it in previous articles, but here’s a brief overview.

Russia has a publicly accessible Probate Registry, which includes all deceased citizens who left any significant assets as inheritance. By comparing it to the named list of losses, we can calculate the probability of inclusion in the registry for each age and social group. Using this coefficient, we can then “convert” the inheritance cases into an actual number of losses—that is, answer the question, “How many actual deaths correspond to the known number of inheritance cases?”

Therefore, this estimate takes into account everyone, not just those who left an inheritance. For groups whose inclusion in the Probate Registry is unlikely (for example, prisoners), we use a higher coefficient of conversion into the deaths estimate, and they are also included in the overall calculation.

As before, this estimate only includes Russian citizens whose deaths have been officially confirmed. But in this update, for the first time, we encountered another group of casualties: those whose deaths were confirmed by the courts—without a body.

In the overall estimate, 261,000 of the 352,000 are “regular” fatalities; their number is calculated using the same method as usual. Another approximately 90,000 are those who were declared dead or missing by court order, or whose deaths were registered with a delay of at least 180 days for other reasons (for example, body swaps or late identification).

We can speak with varying degrees of precision about “regular” and “late” fatalities. More is known about the former; this category is more amenable to analysis. We have less knowledge of the social composition of the latter group. And most importantly: the 90,000 “late” fatalities are an estimate for mid-2025. More recent “belated” fatalities are simply not visible due to the same delay. For at least 180 days after their actual death, they are not reported to the courts or the Probate Registry. More time will have to pass before we can estimate the number of people in this category for the end of 2025.

It’s important to emphasise: this category refers to people who died during combat (not captured or fled the battlefield). They may have been listed as missing in action at some point in the past, but 90,000 is an estimate of the number already recognised as dead.

It would be incorrect to say that we’ve started counting missing persons separately: the difference between these categories lies only in the mechanism by which the death was officially registered. We currently don't know how many military personnel are recognized as missing.

The category of “court-ordered” or “late” deaths is clearly visible both in the Probate Registry and on court websites. In the registry, this group is distinguished by a significant delay between the date of death and the date of its registration. This means that such individuals died, and then, for a long period (at least 180 days), no death certificate was issued for them: either the body was not immediately recovered, or it was not identified, or it was not delivered at all (this is the case with a court-ordered declaration of death or missing).

Claims for declaring servicemen missing or dead began to be filed en masse in July 2024 (Mediazona wrote about this in detail). Based on data in the applicants for these claims, we know for sure that they concern military personnel: the number of such appeals involving ordinary people has remained unchanged, at a rate around 8,000 per year.

At the same time, the number of “late” death registrations has also become noticeable in the Probate Registry. Previously, such cases were rare. Since 2024, we are aware of 52,000 excess probate cases with death registrations exceeding 180 days from the date of death. Applying the average conversion rate of probate cases to actual mortality rates yields 90,000 fatalities.

A similar figure is obtained when calculating court cases. In total, 86,500 “excess” (i.e., exceeding the pre-war level) claims were filed in 2024 and 2025.

In theory, such claims could be duplicated—meaning there would be multiple claims for the same person. This occurs when a person is first declared missing and then later declared dead; or when a judge returns a claim, and the military unit refiles it with clarifications.

A conservative (lower) estimate of the number of unique claims is 61,800. For the lower bound, we subtracted all categories that could, at least in theory, serve as a source of duplicates. These are all claims that were returned, dismissed, or left without consideration, as well as claims in which a person is requested to be declared missing rather than dead (they could potentially later be declared dead by a separate court decision).

To identify the second category, we reviewed 3,800 decisions on missing service members, obtained from a source in the judicial system. 2,600 claims sought to declare a person dead, and 500 sought to declare a person missing. The remaining 700 cases were procedural decisions (error corrections, denials, resubmissions).

Clearly, lawsuits do not cover the entire range of “ultra-late” registrations we see in the Probate Registry. Although such cases were almost nonexistent before the “court claims campaign” (early data shows a rate of about 2.4%), their number could have grown independently of the courts—for example, due to the increased frequency of body exchanges.

Furthermore, lawsuits provide a picture up to mid-2025: although the applications themselves are available until December, they concern people who went missing at least 180 days prior to the filing date (earlier registration is prohibited by law).

Taking this into account, and considering that applying the ”regular” deceased person coefficient to inheritance cases with ultra-late registration yields the same number as the unconservative calculation based on lawsuits (90,000 and 86,500), we decided to estimate this group at the upper limit of lawsuits—90,000 people.