How many people does ICE kill a year on average?
Executive summary
The most robust, peer‑reviewed figure places the long‑term average number of deaths in ICE custody at roughly nine per federal fiscal year — specifically a mean of 8.9 deaths annually from FY2010–2019 — with suicides accounting for about 13% of those deaths (1.2 per year) [1]. Recent reporting shows a sharp spike in 2025, when public notices and press surveys list roughly 30–32 deaths, demonstrating that annual totals can vary widely and that short‑term averages are currently much higher than the prior decade’s mean [2] [3].
1. Historical baseline: a roughly nine‑per‑year average from 2010–2019
A peer‑reviewed retrospective cohort analysis that extracted ICE death reports and FOIA data calculated a mean of 8.9 deaths per fiscal year across 2010–2019, with a standard deviation of 1.7 and an average of 1.2 suicides per year (13% of deaths) [1]. That study used the agency’s own death notices and FOIA library as its source material and expressed rates using the ADP denominator to account for turnover in the detained population [1].
2. Alternative windows and NGO tallies shift the average upward
Advocacy and watchdog analyses covering other multi‑year windows report higher totals: the ACLU reviewed 52 deaths from 2017–2021, a span that would average about 10–11 deaths per year and concluded many were likely preventable [4]. That NGO finding highlights how choice of years and inclusion criteria (for example, whether deaths after release but during agency action are counted) change the computed average [4].
3. 2025 as an outlier year that re‑sets short‑term averages
Multiple news organizations and ICE’s own notices documented an unprecedented rise in deaths in 2025 — roughly 30 deaths reported during that calendar/fiscal year and investigative timelines listing 32 named people — marking the highest annual total since 2004 [2] [3]. That spike, driven in part by a record detained population and several clustered incidents in December, demonstrates that recent short‑term averages (for example, 2018–2025) will be materially higher than the 2010–2019 mean [5] [2].
4. How “ICE kills” is defined matters — reporting rules and gaps
Official tallies count deaths that occur while an individual is in ICE custody and are governed by ICE’s Notification, Review and Reporting Requirements; ICE’s public detainee death reporting explains internal review and notification timelines [6]. Several sources warn of reporting delays, cases buried in press releases, and agency practices (like releasing gravely ill detainees) that can obscure whether a death is reflected in ICE’s custody statistics, meaning raw counts may understate deaths linked to agency custody policies [7] [8].
5. Causes, preventability, and competing narratives
Medical and mental‑health failings are repeatedly cited: the peer‑reviewed literature and recent analyses identify suicides and inadequate medical care as significant contributors, and an ACLU report argued 95% of reviewed deaths from 2017–2021 could likely have been prevented with adequate care [1] [9] [4]. ICE, by contrast, emphasizes systematic reporting, investigations by its Office of Professional Responsibility, and that deaths are a tiny fraction of all people detained — noting in press statements that in‑custody deaths have averaged less than 1% relative to the detained population [6] [5].
6. Bottom line — a defensible answer and its caveats
Using the strongest long‑term academic metric available, ICE detention has averaged about 8.9 deaths per fiscal year (2010–2019), with suicides ~1.2 annually [1]. However, more recent multi‑year NGO counts and 2025’s extraordinary toll (approximately 30–32 reported deaths) push short‑term averages substantially higher and underscore that annual death totals are sensitive to policy shifts, detention population size, reporting practices, and whether advocates’ broader inclusion criteria are applied [4] [2] [3]. Sources differ on cause and preventability — NGOs and academic studies emphasize systemic neglect and preventability [4] [9], while ICE stresses formal reporting procedures and relative rates [6] [5]. The available public evidence therefore supports a long‑term average near nine deaths per year but also documents a marked and concerning rise in recent years that changes the near‑term average and public policy stakes [1] [2] [3].