Assessing Russian Casualties

Part I: Gathering Data

Most of my time over the past couple weeks has been spent preparing an in-depth estimate of Russian casualties and equipment losses for Neue Zürcher Zeitung. This estimate is being done for each unit for each week of the war and I am aiming for completion (Part II) in the next two weeks or so. My numbers are not finalized yet, but the following outlines some of my thoughts thus far.

Here are some of the notable existing recent Western and Ukrainian estimates:

• 10 November 2022: US General Milley stated 100,000 Russian casualties. Last week, he updated the number to “well over” 100,000.¹ Casualties in this sense would included both dead and wounded.

• 22 January 2023: Norwegian Chief of Defence Eirik Kristoffersen stated 180,000 dead or wounded for Russian losses.² Notably this is a bit of an outlier from General Milley’s statement.

• 24 January 2023: Ukrainian General Staff daily update puts the number at 122,170 “liquidated” or “eliminated” (the word used is ліквідовано).³ I have not seen a definitive answer on if this estimate includes wounded or not. The word itself would seem to imply killed. If it does include wounded however, it would be broadly similar to General Milley’s statements. It might also be that General Milley derived his numbers from these public Ukrainian estimates rather than from an independent assessment from the US intelligence community.

The frustrating thing about the above estimates is the lack of published methodology (likely due to classification) and the lack of clarity on what these number include. It is unclear if these numbers include the self-declared separatist forces of the DPR and LPR. Based on the above numbers, the Western estimates of Russian forces killed in action probably stand at around 40,000, assuming they are using an approximate ratio of three wounded for every fatality. The three to wound wounded/killed ratio is highly dependent on an armed forces reporting procedures for wounded personnel.

There are a handful of official statements from the Russian side on their own casualties.

• 23 December 2022: DPR press release of 4,176 military personnel killed in 2022. 13 of these occurred prior to February 25, making the total for the full-scale war 4,163. ⁴ This number is updated periodically.

• 21 September 2022: Sergei Shoigu states 5,937 Russian dead.⁵ I have not seen an official Russian Armed Forces statement on their losses since then.

When Shoigu made his statement, the DPR number stood at 3,318. Assuming LPR losses “might” be similar to the DPR, it “might” be assumed that Russian regular forces losses are approximately equal to both the LPR and DPR put together. “Might” is doing a lot of heavy lifting here (much like the arrows around Bakhmut on the typical pro-Russian map). It could also be true that DPR is being slightly more forthcoming with their losses that the regular Russian military. The main point here is there is obviously a wide gap between the western estimates and what the Russians themselves state their losses are.

Some of this can be filled by open-source investigations. The most important of these is the joint BBC/Mediazona collection of information on Russian military funerals and gravesites.⁶ Their number stands at 11,662 currently. This number can be subdivided by branch and oblast in most instances. This number by no means is complete with the BBC believing the list contains 40 to 60 percent of the number that have been buried in Russia.⁷ New entries are being added each day. For the first month of the war, there are currently 1,917 confirmed deaths. Wayback Machine shows Mediazona and BBC found 1,782 deaths for the same time period on 1 October, highlighting the continuing backlog.

We should also be willing to accept at least the possibility the western and Ukrainian estimates are too high. If one believes that the BBC/Mediazona database represents about half of all killed in action, this would result in a determination of about 25,000 dead from the Russian Armed Forces (this would also include the National Guard and some of Wagner’s losses, but not include the separatist forces). If separatists are included, the number of killed in action might approach or surpass the 40,000 mark as previously stated. I do not have a firm opinion on the numbers yet, but overestimations of enemy losses are common in major conflicts. The Russians too are almost certainly overestimating the amount of losses they are inflicting on the Ukrainians. The pro-Russian side will frequently take a small amount of footage or a quote from a Ukrainian official out of context and immediately jump to overarching conclusions about the front as a whole.

Returning to the task of assessing Russian losses, we also have the work of Ragnar Gunmundson and Simon-Pierre Lussier. They have been able to independently determine the dates of death of many Russian officers.⁸ Ragnar kindly provided the underlying officer data prior to the depicted timeframe. You can see the spikes in confirmed officer deaths (in red) appear to correspond to spikes in the Ukrainian estimates (in white).

This indicates the Ukrainian estimates probably have at least some basis in reality and might be interesting to explore from someone more mathematically inclined than I. If there is an error in the Ukrainian estimates, it is likely in the overall number of assessed Russian losses rather than the broader chronological patterns. As the graph gets closer to the present, the correlation between officer deaths and Ukrainian numbers decreases, but this is perhaps due to the inherent delay in reporting, as was the case with BBC/Mediazona’s data.

Personally, I was able to determine the unit for about 1,200 Russian killed in action so far via developing a couple of webscrappers and also manually reading various pro-Russian blog sites. Another additional 300 were determined by matching oblast with branch of service from the BBC/Mediazona data. For example, a VDV member from Pskov Oblast is almost certainly from the 76th Guards Air Assault Division. This number will likely go up given more time and research on my part.

We also have the famous Oryx list. Russo-Ukrainian Warspotting has taken the Oryx list and categorized losses by unit where possible.⁹ The identification of equipment to unit is generally accomplished through painted tactical symbols on each vehicle. The dates given should be regarded as the vehicle was observed on social media instead of the precise date the vehicle was lost or destroyed so their is some margin of error.

Below are three of the previously discussed metrics displayed graphically by week. Officer deaths in orange from Ragnar Gunmundson and Simon-Pierre Lussier, all confirmed deaths associated with specific deaths from me in blue, and armored fighting vehicle (AFV) losses associated with specific Russian units from Warspotting in grey. The positive correlations are as follows: officer deaths and unit deaths 0.86 (high), officer deaths and AFV losses 0.62 (moderate), and unit deaths and AFV losses 0.53 (moderate). The large spike in vehicle loses in mid to late September corresponds to the Kharkiv counteroffensive.

The correlation between armored vehicles and casualties has been studied in the west previously, notably in Colonel Trevor Dupuy works in the late 1980s. ¹⁰ We can begin to start to validate some of his models by looking at captured documents from the 1st Guards Tank Army while referencing the Oryx data from Warspotting.¹¹ Below is a very simplified (and subject to change) order of battle for the units involved. I have assumed BTG's of 40 infantry fighting vehicles and tanks each. For simplification purposes, I have designated AFV's to only include tanks and infantry fighting vehicles. For 2nd MRD, IFV's include BTR-80's which normally would not be classified as an IFV by OSCE standards. BTG's would of course have more vehicles than tanks and IFV's; these were selected as they were the most widely and accurately tracked. Personnel estimates are simplified for illustrative purposes and represent both personnel within the BTG and from supporting brigade and divisional units not within the BTG's.

Below are each of those subunits estimated losses for the initial three weeks of the war. Vehicle losses come via Warspotting and was similar to vehicle losses reported in the captured Russian document. Derived personnel casualty rate was determined from (daily AFV losses) / (total AFV count x 6) for each unit for each week. This equation was developed by Dupuy by analyzing data from twentieth century conflicts involving armor heavy formations.¹² Somewhat simplifying this process, I extrapolated the resulting daily casualty projection and multiplied by seven to get a weekly result. The end product is 400 assessed casualties. This eerily almost exactly matches the total in the captured Russian document of 408 (61 killed, 44 missing, 207 wounded, and 96 captured).

This same process might be able to give us casualty estimates on other Russian mechanized forces, during the early stages of the war in particular. The breakdown of killed, missing, wounded, or captured would likely be highly variable however. 1st GTA lost many of its vehicles through breakdown rather than fighting, resulting in more missing and captured than might otherwise be the case. We are also unlikely to have near-complete vehicle loss lists for all units. In these cases, I would probably need to rely on some combination of confirmed deaths and Ukrainian estimates and reports.

The full estimates of Russian losses will be done in several weeks, likely around 24 February. Like all wartime related data, it is meant to be preliminary rather than authoritative. As more information comes out, the data will undoubtedly change. I probably will not have much time until mid-February, so I might not have any additional posts till then.

1. Caitlin Doornbos, “Russia has ‘significantly’ more than 100,000 casualties in Ukraine: US Gen. Milley,” New York Post, 20 January, 2023, https://nypost.com/2023/01/20/russia-has-significantly-more-than-100000-casualties-in-ukraine-milley/.

2. “Russia taken 180,000 dead or wounded in Ukraine: Norwegian army,” France 24, 22 January, 2023, https://www.france24.com/en/live-news/20230122-russia-taken-180-000-dead-or-wounded-in-ukraine-norwegian-army.

3. Ukrainian General Staff Update, 24 January, 2023, https://www.facebook.com/GeneralStaff.ua/posts/pfbid0j4qrUUCkaHY12UCxfuURV7ntzjj5BuDAkWsj6eY58KpStooWCJ5eHU8CfDxztfmMl?__cft__[0]=AZVX8mfatlXG8oPqz3ZA6LtY7OVwY3UhZDhFX9ZLWzkibYxmoXhQMkSz5vGW6mP3XiOPu6Rd4J6cBrFLuKZ12VcoisUMPpAlEY6ZNxuLUxkQjuvLtuoiFCKgSAPS3qiPH3hvazAYLog_hpIf4wicnP7OrfhfwwNR-0rGSn6bj_LJ-CvrgUygblxJVo2I_FvmLM0sA9lahRzu60GNWOFJd9nP&__tn__=%2CO%2CP-R.

4. “Overview of the current social and humanitarian situation in the territory of the Donetsk People’s Republic as a result of hostilities,” Human Rights Ombudsman in the Donetsk People’s Republic, 23 December, 2022, https://eng.ombudsman-dnr.ru/the-overview-of-the-current-social-and-humanitarian-situation-in-the-territory-of-the-donetsk-peoples-republic-as-a-result-of-hostilities-in-the-period-17-and-23-december-2022/.

5. “Defence Minister Says 5,937 Russian Soldiers Killed In Ukraine,” 21 September, 2022, https://www.barrons.com/news/defence-minister-says-5-937-russian-soldiers-killed-in-ukraine-01663743607?refsec=topics_afp-news.

6. “Russian casualties in Ukraine. Mediazona count, updated,” 17 January, 2023, https://en.zona.media/article/2022/05/20/casualties_eng.

7. Olga Ivshina, “10,000 names: what is known about Russia's losses in Ukraine by December,” 9 December, 2022, https://www.bbc.com/russian/features-63917502.

8. Ragnar Gudmundsson and Simon-Pierre Lussier, “Tracking Russia's losses in Ukraine,” https://datastudio.google.com/reporting/dfbcec47-7b01-400e-ab21-de8eb98c8f3a/page/p_ykn9uao30c?s=tJipopktA34.

9. “Russo-Ukrainian Warspotting,” https://ukr.warspotting.net/.

10. See Trevor Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War, 1990 and Shawn Woodford, “Tank Loss Rates in Combat: Then and Now,” 25 October, 2016, http://www.dupuyinstitute.org/blog/2016/10/25/tank-loss-rates-in-combat-then-and-now/#:~:text=In%20combat%20between%20two%20division,10%20for%20the%20losing%20side.

11. “By mid-March, about a hundred Russian tankers had surrendered,” Ukrainian Ministry of Defense Directorate of Intelligence, 16 May, 2022, https://gur.gov.ua/content/do-seredyny-bereznia-zdalysia-v-polon-blyzko-sta-rosiiskykh-tankistiv.html.

12. Dupuy, Attrition, p. 112.