
Since 2023, Meduza, together with our colleagues at Mediazona, has been analyzing Russia’s National Probate Registry to estimate the total number of Russian soldiers killed in the current war and track the dynamics of these losses over time.
The most reliable source of data on these losses is confirmed casualty lists that use obituaries, social media posts, and other open sources the confirm individual soldiers’ deaths. Mediazona compiles these lists together with BBC News Russian and a team of volunteers. However, they’re always incomplete, and it’s impossible to know how many fallen soldiers have not yet been included.
This is where the National Probate Registry comes in. We can use this registry as a second independent data source, allowing us to assess both the completeness of the confirmed casualty lists and Russia’s overall number of losses. That completeness level is constantly changing: it varies by year and by age cohort of the deceased. This means there’s no universal coefficient that could be applied to a confirmed list to estimate total losses.
We described our analysis of the probate registry in detail in the first article in this series, so we won’t go into the full methodology here. The analysis — especially collecting the raw registry data (tens of millions of probate cases) — is time-consuming, but the process itself is fairly straightforward.
The key insight from our latest analysis of National Probate Registry data is a sharp rise in losses, which seem to reach a new level of intensity each year. Now, six months after our previous estimate, we can confirm that this increase persisted in 2024, with losses hitting record highs. During some periods, the Russian Armed Forces lost more than 2,000 men per week, and overall, according to the National Inheritance Registry, about 93,000 Russian servicemen were killed last year — almost twice as many as in 2023, when the death toll was roughly 50,000.
The biggest challenge in producing timely estimates (aside from collecting the data itself) is the method’s built-in “blind spot”: it can’t reliably account for losses in the most recent six months. The main reason for this is that relatives have at least 180 days after a death to open a probate case, which means the latest half-year of data in the registry is always incomplete. This gap can be partially corrected by using historical probabilities of how soon probate cases are opened after a death, which helps reconstruct the picture. But this requires additional assumptions and leaves room for error, since it depends on families’ behavior around probate cases remaining consistent on average.
To address this, we’ve adopted a different approach in our latest analysis: we only use registry data once it’s complete. For the most recent six months we use a provisional estimate based on the total number of deaths in the named casualty lists, the rate at which these lists grow, and the ratio between that number and the number of people in the same cohorts on the same dates in the lists. In other words, we use a predictive model.
This model gives fairly reliable results for the overall death toll, but the week-by-week distribution can be heavily skewed — depending on how quickly Mediazona and BBC volunteers process potential entries, whether new leaks of casualty data appear, and other random factors. On top of that, the named lists also lag behind, and new fatalities don’t appear there immediately, which creates a steep drop in the charts for recent months. It’s important to remember that this drop doesn’t mean losses have decreased — it only reflects the incompleteness of the lists.
Luckily, it’s possible to gauge the relative intensity of losses using the same inheritance data. To do this, we can look at the ratio of probate cases for men versus women, which clearly shows that the increase in deaths at the end of 2024 was real. We can’t say with confidence whether losses at that point actually crossed the 3,000-per-week mark or were slightly lower (the accuracy of the calculation drops sharply in this period), but the growth is confirmed by the simple male-to-female case ratio in the registry — and it aligns with a small spike in deaths on the named lists (which are also still highly incomplete at this point).
Taking all the uncertainties of the predictive model into account, we estimate that by the end of summer 2025, the total number of Russian military personnel killed in the war was around 220,000.
Another factor, whose effect is still hard to measure and which has only recently started influencing casualty counts, is the courts officially declaring missing soldiers dead. An analysis of civil cases filed in district and garrison courts under the category “Declaration of a citizen as missing or deceased” shows that these cases surged sharply in the second half of 2024.
By mid-2025, the number of these cases had reached 2,000 per week. In other words, courts are now “adding” to the death toll almost as much as the fighting itself.
Of course, not everyone listed as missing in court cases is a soldier killed in Ukraine. Court rulings usually redact names, which makes it impossible in most cases to match the rulings to the names of missing servicemen (which are usually found in social media groups for families searching for their lost relatives). Sometimes a court decision makes it obvious that the person was a soldier, but these cases are the exception, not the rule.
Still, the rise in missing persons by late 2024 is also visible in the National Probate Registry. One way to see it is by looking at the number of inheritance cases opened for people where a significant gap exists between the date of death and the official death certificate. Such gaps — weeks, months, or even years — did occur before the war. But when we adjust these delays to pre-war averages, a pattern emerges: it was in 2024, not right after the war began, that the number of inheritance cases involving “lost” decedents started to climb. This increase mirrors the trend in court cases.
The reason we can assume that these cases are related to the war is that the growth is observed exclusively among men. In the same chart for women, there’s no change at all — neither after the war began, nor in recent months.
It’s worth noting that the previous chart grouped cases by the date the inheritance case was opened — showing the growth in new filings and how the share of missing persons within them has increased. But these people may have died long before the case was opened. On the casualty charts we showed earlier — where cases are grouped by death date — the same individuals would appear much earlier, in the year and week when they died (or when the court ruled that they died).
Courts handle death dates in different ways, and so far we know only a few examples of how this works for missing servicemen:
Still, a significant portion — and likely the majority — of such cases are assigned to earlier years, usually close to the actual time of death. This is clear in the next chart, which shows the same inheritance cases as before, but distributed by date of death.
It’s clear that courts tend to place these deaths (ones with a long gap between the actual death and the official death certificate) mostly “in the past.” From the standpoint of tracking loss dynamics, that’s a good thing — it introduces less distortion than if the date of death were simply set to the day the court issued its ruling. If that were the case, we’d see a surge in mortality only now, when these court decisions started coming through.
Courts may have contributed slightly to the increase in reported deaths over the past six months, but the effect seems small — otherwise, it would show up clearly in National Probate Registry data. For now, we don’t know how many missing soldiers are in the registry, whether this group can be singled out, or how the growing wave of court cases might distort the broader picture of casualty trends in the future. All of this will require fresh data that covers the past six months and a dedicated analysis.
Meduza’s Razbor (“Explainers”) team