Home / Deaths
Deaths during police encounters
37,956 deaths recorded between 2000 and 2026, attached to the agencies involved. Police Census publishes counts only and does not name the people who died.
Read this before the table. A death recorded here is not a finding of wrongdoing, and this page is not a ranking of which police forces are dangerous. Many of these deaths involved armed confrontations; some involved none. The figures record what happened, not who was at fault.
There is no comprehensive federal record of deaths during police encounters. The FBI’s own use-of-force collection is voluntary and reports at state level only. These records are compiled by two independent research projects, Mapping Police Violence and Fatal Encounters, described under Source below. They are not official statistics.
6,815 of the 37,956 records could not be attached to an agency in the FBI’s current register — for Mapping Police Violence records, usually because the agency identifier they carry has since been retired; for Fatal Encounters records, because the agency name did not resolve to exactly one agency. Those deaths are counted in the national total above but do not appear on any agency page.
Highest death rate per 100,000 people policed, 2016–2025
Raw counts are not a ranking on this site: the largest forces will always top a raw list because they police the most people. This table counts the deaths recorded for each agency from 1 January 2016 to 31 December 2025 and divides by the population that agency reports it serves. It is a rate over the whole 10 years, not an annual rate: a figure of 4.1 means 4.1 deaths in 10 years for every 100,000 people, about 0.41 a year.
Why the window is fixed. An agency’s deaths over every year on record, divided by a single year’s population, compares agencies over spans of different length — for example, an agency with 22 years of records would rank below an otherwise identical one with 27, for a reason that has nothing to do with policing. Counting every agency over the same 10 years is what makes one row comparable with another.
2026 is left out on purpose. It is a partial year (records run to 18 August 2026), so it is excluded for every agency rather than counted for some. Only agencies policing at least 100,000 people with at least 5 deaths in the window are ranked — below that, one death produces an arithmetically true but meaningless rate. The 5-death minimum was kept when the window shortened: shorter windows give smaller counts and noisier rates, so it is no less needed.
Per 100,000 = deaths 2016–2025 × 100,000 ÷ population. Each death is counted once: records that appear in both sources are excluded as duplicates. Population is the figure each agency itself reports to the FBI as the population it serves, from its 2025 return (its 2024 return for Kansas City Police Department, KS, which had not filed for 2025). It is a single year’s figure applied to 10 years, so a fast-growing or shrinking jurisdiction’s rate is somewhat understated or overstated. An agency with no 2024 or 2025 return is not ranked. Jurisdictions overlap, so a death in a city may involve a county or state agency as well as the city force.
What this ranking still cannot support. It is not a finding of wrongdoing by any agency, and it does not show which forces are more dangerous. It is not a trend: the two sources do not cover the same years, and 2021 is the last year Fatal Encounters covers, so 2016–2021 draws on both sources and later years on Mapping Police Violence alone. That applies to every agency equally, which is why agencies can be compared with each other, but it means the early and late halves of the window are not like for like. Deaths that could not be attached to an agency (see above) appear in no row, so an agency whose records were hard to match is undercounted. With counts this small, a difference of a few deaths can reorder neighbouring rows.
Source
This page combines two sources. They cover different years, and neither is credited with a figure it did not record.
- Mapping Police Violence (MPV): 15,550 records, 2013–2026. It is the only one of the two covering 2022 onward. Records are attached to agencies using the agency identifier MPV publishes alongside each incident.
- Fatal Encounters: 31,497 records, 2000–2021. The project states: “we let anyone use the data for any reason for free”. It publishes the agency as free text rather than as an identifier, so a record is attached to an agency only when the name resolves to exactly one agency in that state; a record naming several agencies, or none that resolve, is counted in the national total but appears on no agency page.
Why the total is 37,956 and not 47,047. The two sources overlap in 2013–2021, so the same death can appear in both. 9,091 Fatal Encounters records are excluded as duplicates of MPV records, so that each death is counted once. 15,550 + 31,497 − 9,091 = 37,956. Until 11 October 2026 this site did not make that exclusion and published 47,047— the duplicates counted twice, and the per-capita ranking above was in the wrong order as a result. No records were lost: the excluded records are kept, flagged, and are not counted. See the published correction.
A match requires the same state, the exact incident date and the same first and last name (the name is used only to match, and is never displayed). A death recorded with a different date or spelling in the two sources would not be matched, so the total may still count a small number of deaths twice. Because the sources cover different years — 2000–2012 from Fatal Encounters alone, 2022 onward from MPV alone — counts spanning those years are not a like-for-like series. Counts as at 11 October 2026.
Causes and official dispositions are reproduced as published and are never paraphrased. Mapping Police Violence data is displayed with attribution to its compiler; the written scope of this site’s right to display it is being confirmed (note dated 11 October 2026).