Counting bodies, casting doubts Independent Russian analysts dispute the reliability of alleged Russian casualties leaked by Ukraine’s military intelligence
This week, researchers at the independent news website Mediazona and the OSINT project Conflict Intelligence Team disputed the reliability of alleged Russian military records leaked by Ukraine’s military intelligence. The data — showing nearly 87,000 Russian soldiers killed in the first eight months of 2025 — was published by “I Want to Live,” a project run by the Ukrainian Defense Ministry. Meduza explains the two outlets’ suspicions about the numbers.
Conflict Intelligence Team (CIT) analyzed the alleged military loss records for statistical signs of manipulation using forensic methods similar to those employed in election-fraud investigations. As CIT explains, “The approach is based on findings from psychological research showing that when people are asked to generate random numbers, they favor certain digits unconsciously, which becomes evident in the data.”
Researchers found several anomalies.
Specifically, CIT found the distribution of last digits among the reported figures “deviates sharply from a true random pattern,” with “too many ones and twos and not enough sixes and sevens.” This was quantitatively confirmed by Pearson’s chi-squared test, which measures how closely observed results match an expected statistical distribution. CIT concluded that the chance that the leaked data occurred randomly was only 1.3 percent.
When CIT separated the numbers into primary (raw, unit-level) and aggregated (sums, totals) categories, the anomalies were even more pronounced in the primary data. There, the probability of seeing such a pattern by chance fell to just 0.4 percent. “In other words, it looks like the document’s author filled in figures for each unit separately before adding them up into aggregate totals. That process caused the distribution of final digits — apart from zeros and fives — to even out and become uniform,” CIT concluded.
In Mediazona’s biweekly update on Russian battlefield losses, editor Dmitry Treshchanin echoed CIT’s concerns about the data from Ukrainian intelligence. He also acknowledged that the Russian military itself could be responsible for manipulating these records. “Actually, it’s hard to imagine a Russian officer who doesn’t tamper with the records,” Treshchanin explained. He also argued that “glaring omissions” in the leaked records (such as missing figures on desertions) cast doubt on the authenticity of the leak.
Other irregularities are apparent, particularly from a military bureaucratic perspective: the documents lack a clear recipient and an obvious purpose. Neither is there any indication as to who produced the records, when they were submitted, or where they were filed.
At the same time, Treshchanin emphasized that Mediazona and its partners are not currently able to refute or corroborate the leaked claim that nearly 87,000 Russian soldiers were killed in Ukraine in the first eight months of 2025. “We’ll only have enough confidence to assess these numbers in 2026, and not at the start of the year, but more likely by next fall,” he explained. This is because Mediazona restricts its count of killed soldiers to published obituaries and its estimates of deaths to public probate records.