The Autopsy / Case File · Violence & Homicide
Nobody Owns the Count
Ask how many Americans die every year in a police encounter and the honest answer is: nobody knows for certain, and the agency legally responsible for finding out has spent more than a decade not fully doing it. A 1994 law told the Justice Department to collect this data. A 2013 law tried again, with penalties attached. Neither has produced a complete count.
The Death in Custody Reporting Act of 2013 authorized the Attorney General to withhold up to 10 percent of a state's federal justice-assistance grant funding if it failed to report deaths of people in police custody or during arrest. A 2018 Justice Department Office of Inspector General review found the Department hadn't even begun collecting state-level data by the statutory 2016 deadline — a four-year gap — and that its own prior arrest-related-death program had captured only about half of the law enforcement homicides independent researchers could identify by other means. The most complete public substitute, a Washington Post database that had run since 2015, quietly stopped updating on January 1, 2025.
Join the discussion →
American law enforcement operates roughly 18,000 separate agencies, and there is still no single federal system that reliably counts how many people die each year in an arrest, a police shooting, a struggle, or in custody afterward. That absence isn't for lack of legislative effort — Congress has passed two separate laws specifically ordering the Justice Department to collect this data, in 1994 and again in 2013 — but both have run into the same wall: reporting from state and local police is voluntary in practice even when it's mandatory on paper, and the federal government has never built an enforcement mechanism with real teeth behind it.
Two Laws, Three Decades, One Persistent Gap
The first attempt came in the Violent Crime Control and Law Enforcement Act of 1994, which directed the Attorney General to "acquire data about the use of excessive force by law enforcement officers" and publish annual national summaries. It produced little usable data and no sustained public reporting system. The Bureau of Justice Statistics did run an Arrest-Related Deaths program from 2003 through 2012, a national census attempting to track deaths during the arrest process, including demographic details, cause of death, and weapon involvement — but BJS's own later research, comparing that program's tallies against the FBI's Supplementary Homicide Reports for the same 2003-2011 period, found that each source independently captured only about half of the law enforcement homicides the two together could identify. Neither dataset alone came close to a real national count, and the program itself lapsed.
Congress tried again with the Death in Custody Reporting Act of 2013, signed into law in December 2014, which required state and federal law enforcement agencies to report to the Attorney General the death of anyone detained, arrested, being arrested, being transported to custody, or incarcerated — and, notably, gave the law actual enforcement teeth: the Attorney General could withhold up to 10 percent of a noncompliant state's Edward Byrne Memorial Justice Assistance Grant funding. The statute set a deadline: state-level data collection was supposed to begin by fiscal year 2016.
What the Inspector General Actually Found
The Justice Department's own Office of Inspector General reviewed DCRA's implementation in 2018 and documented a program badly behind schedule and structurally uncertain of its own scope. The Department had not yet collected state-level arrest-related death data as of the review, despite the statutory requirement to have started by fiscal year 2016 — a report that put the actual expected start date closer to fiscal year 2020, a roughly four-year delay. On the federal side, 104 federal agencies had submitted the required reports, but the OIG found the Department lacked a complete inventory of which federal agencies even had law enforcement authority in the first place, estimating that somewhere between three and thirteen agencies with reporting obligations may never have reported at all. The Federal Bureau of Prisons, meanwhile, consistently failed to provide a statutorily required data field: the time of death.
The OIG review also flagged something structurally revealing: three separate components within the Department had each independently begun planning overlapping data-collection efforts, raising the prospect of duplicative demands on the same reporting agencies, wasted federal resources, and — the OIG's own phrase — a real risk of "compromised data quality" from confusion alone. And a separate requirement buried in the same 2013 law, that the Department deliver a study to Congress within two years of enactment on how DCRA data could actually be used to reduce in-custody deaths, had never been produced at all as of the OIG's review, four years past its own deadline.
Each of two federal data sources, run by the same federal government eight years apart, independently captured about half of the deaths the other could identify. Neither agency knew what the other was missing.
The Administrative Handoff Problem
By 2019, administration of DCRA's data collection had shifted from the Bureau of Justice Statistics to the Bureau of Justice Assistance — the agency responsible for actually holding the ten-percent-of-JAG-funding penalty over noncompliant states, rather than the agency built for pure statistical research and public reporting. That handoff reflects an underlying tension DCRA never fully resolved: BJS, the statistical arm, has the methodological credibility to build a nationally representative dataset, but only BJA holds the compliance stick — and the version of the program that actually threatens funding penalties has published far less detailed, research-grade public data than BJS's earlier statistical efforts did. A state facing a modest grant-funding penalty for incomplete reporting has a real incentive to report something; it has much less incentive to report something detailed enough for outside researchers to actually analyze use-of-force patterns by circumstance, weapon, or demographic group — which is precisely the level of detail the original Arrest-Related Deaths program's design was aiming for before it lapsed.
The Database That Filled the Gap, Then Stopped
Into that federal vacuum stepped journalism. The Washington Post's Fatal Force database, launched in the aftermath of the 2014 killing of Michael Brown in Ferguson, Missouri, tracked every fatal police shooting nationwide starting in 2015 using local news coverage, public records, and law enforcement reports — earning Pulitzer, Peabody, and Polk recognition along the way and becoming the dataset researchers, journalists, and policymakers most often cited as the closest thing to a real national count, precisely because the federal government hadn't produced one. On January 1, 2025, the Post quietly stopped updating it. The archived data for 2015 through 2024 remains accessible, but the database that had effectively substituted for a federal system for a decade no longer functions as a live one — leaving independent trackers, state-level efforts like Wisconsin's use-of-force reporting (itself hampered, per recent state-level reporting, by numbers that are difficult to compare agency to agency), and whatever DCRA eventually produces as the remaining options. None of them, individually or combined, currently amounts to what Congress ordered twice: one complete, consistently defined national count.
And Then You're Dead
A death during a police encounter is one of the most consequential things a government can be asked to count honestly, and it's also one of the events American government has spent three decades and two separate acts of Congress failing to count completely. The gap isn't a mystery in need of new legislation. It's an old law, still not fully implemented, next to a database that used to fill in for it and now doesn't.
That's it. That's the whole thing.
Sources
- The Death in Custody Reporting Act of 2013, its 10% JAG-funding withholding penalty, and its FY2016 state-data-collection deadline — Public Law 113–242, "Death in Custody Reporting Act of 2013," December 18, 2014; Congressional Research Service Report R47559
- The DOJ Office of Inspector General's 2018 findings on DCRA implementation delays, the incomplete federal agency inventory, the Bureau of Prisons' missing time-of-death data, duplicative data-collection planning, and the unproduced congressional study — "Review of the Department of Justice's Implementation of the Death in Custody Reporting Act of 2013," U.S. Department of Justice Office of the Inspector General, 2018
- The Bureau of Justice Statistics' Arrest-Related Deaths program (2003–2012), the 1994 Violent Crime Control and Law Enforcement Act reporting mandate, and BJS's finding that FBI Supplementary Homicide Reports and ARD data each independently captured roughly half of estimated law enforcement homicides, 2003–2011 — Bureau of Justice Statistics, "Use of Force" topic overview, U.S. Department of Justice
- The Bureau of Justice Assistance's 2019 assumption of DCRA data-collection administration from BJS — Bureau of Justice Assistance, Death in Custody Reporting Act program materials
- The Washington Post's Fatal Force database, its 2015 launch following the killing of Michael Brown, and its cessation of updates as of January 1, 2025 — "Fatal police shootings are increasing, but fewer are being reported," The Washington Post; "The Washington Post's now-shuttered Fatal Force database," American Experiment
- State-level use-of-force reporting comparability challenges — "Wisconsin tracks police use of force, but the numbers are hard to compare," Wisconsin Watch, 2026
- Photo: "Police car with emergency lights on," Scott Davidson, CC BY 2.0, via Wikimedia Commons
Comments
Loading…