Death rate by fiscal year
Death rate = deaths ÷ average daily population × 100,000. Fiscal year runs Oct 1 – Sep 30.
This is all-cause mortality. For the suicide-specific surge — a different and narrower signal — see the next chart.
Suicides by fiscal year — the surge
A subset of the all-cause chart above, and our own classification: we read every published ICE Detainee Death Report and coded the manner from how each report describes the death (e.g. "ligature around his neck", "hanging from a bed sheet", "cut-down tool" → suicide). ICE prints no manner field. Reports exist only from FY2018 on, so the series starts there. FY2020 is the COVID-era cluster; FY2026 = 7 is the highest in the published-report era, with months still to run.
Across Jan 2025 – Apr 2026 we classify 10 deaths as suicide (3 in FY2025, 7 in FY2026), 3 as homicide (two men shot at a Dallas facility; one death ruled a homicide by a medical examiner), and 7 as undetermined — the report states no method or cause, often pending autopsy. Filter the records table by Manner of death → suicide, then click any ICE report ↗ link to read the primary document.
Year summary
Click a column header to sort. Click a year to filter the records table below.
| Fiscal year | Deaths | Avg daily population | Death rate /100k |
|---|
Deaths by country of origin
All — deaths in ICE custody, Jan 23 2025 – May 22 2026, by the detainee's country of origin. Click a column header to sort.
| Country of origin | Deaths | Share of total |
|---|
Before you publish: these counts cover all 51 deaths. Primary source: each death's official ICE Detainee Death Report, linked per row in the records table below — country read from the report's "Country of Citizenship" field. Index: ice.gov/detain/detainee-death-reporting ↗. China (3) is confirmed from the reports for Ge, Wong and Xie.
Individual death records
One row per person. "Not reported" means the field was not published for that death — never a guess.
| Date | FY | Name | Country | Facility | Manner | Source |
|---|
CSV format to load the full ICE list (header row required):
date,fy,name,country,facility,manner,ddr
Story leads
Generated from the working dataset above (FY2004–FY2025, 237 deaths total). Numbers are approximate pre-2018 and preliminary for FY2025. Angles marked LEAD are the recommended starting point.
| # | Angle | Key finding — count & rate | Why it matters | Biggest confound to defend | Top reporting steps |
|---|---|---|---|---|---|
| 1LEAD | Deadliest year by the count — but is it by the rate? | FY2025: 25 deaths, rate 50.0/100k (ADP 50,000). All-time high count, but ~3rd-highest rate behind FY2004 (131) & FY2005 (97). | Resolves the "deadliest year" claim everyone is making — honestly. | Pre-2018 data is unreliable; if early counts are inflated, FY2025 is the worst rate too. | Re-derive FY2004–07 counts from ICE FOIA death list; lock ADP source; interview detention-mortality epidemiologists. |
| 2 | More people locked up, more people dying | FY2021 (5, ADP 19,254) → FY2025 (25, 50,000). Both axes up. | Ties deaths to the FY2024–25 detention surge. | Rising count from rising population ≠ worse care; the rate isolates that. | Pull monthly ADP; test if rate rose faster than population; OIG reports on surge-era facilities. |
| 3 | The COVID spike the count hides | FY2020: 21 deaths, rate 62.3/100k — 2nd-highest rate, while ADP fell to 33,724. | Pandemic-era care failure a raw count understates. | COVID is a one-off external shock. | Cross-check death dates vs facility outbreaks; Otay Mesa/Stewart records; DHS OIG COVID reports. |
| 4 | Which facilities keep coming up? | Facility-level clustering (load full record list). | Names specific operators/contractors — accountability. | Bigger facilities have more deaths by size — normalize per-facility ADP. | Load full ICE list; rank by deaths and by deaths/bed; map to CoreCivic/GEO. |
| 5 | Suicide as a manner of death | Manner breakdown (load full record list). | Signals mental-health-care and solitary failures. | Small counts → volatile; "undetermined" hides cases. | Filter manner=suicide; FOIA solitary logs & ME reports; interview detention-MH experts. |
| 6 | Country-of-origin pattern | Country breakdown (load full record list). | Diplomatic / consular-notification accountability. | Reflects who's detained, not differential risk — normalize vs population mix. | Filter by country; compare to detained-population nationality mix. |
| 7 | Small numbers, big swings (methods/watchdog) | FY2021–23 sat at 4–5/yr; one extra death moves the rate sharply. | Inoculates readers (and you) against rate volatility. | This is the confound — make it the subject. | Run confidence intervals on annual rates; consult a biostatistician. |
| 8 | Cause of death: medical care under scrutiny | Cause field (load full record list) — cardiac/medical vs preventable. | Quality-of-care and medical-staffing cuts. | "Natural" can still mean neglected. | Pull Detainee Death Reviews; compare cause to prior medical-grievance records. |
Five FOIA requests to file this week:
- Complete in-custody death list FY2018–FY2026 (name, date, facility, country, cause, manner) — the structured source to replace this page's placeholder rows.
- Detainee Death Reviews (DDRs) for every FY2024–FY2025 death — internal cause/care narratives.
- Monthly average daily population by facility, FY2020–FY2026 — needed for per-facility death rates (angle 4).
- DHS OIG inspection reports & corrective-action plans for the top facilities surfaced in angle 4.
- Medical-staffing contracts and vacancy reports for those facilities — the care-failure paper trail behind angle 8.
Methodology & sources
- Death rate = deaths in custody ÷ average daily population (ADP) × 100,000, by federal fiscal year (Oct 1 – Sep 30). ADP is the average number of people detained on any given day that year — not total book-ins. A rate based on book-ins would be far smaller and is a different statistic.
- Death counts: ICE Detainee Death Reporting page and ICE/DHS FOIA death lists; for FY2018+ ICE is required to publish each in-custody death within 90 days.
- Average daily population: ICE ERO statistics, DHS budget justifications, and the Vera Institute ICE Detention Trends dashboard (daily population since FY2009).
- Manner of death: classified by us from each ICE Detainee Death Report's own death-event narrative (ICE prints no manner field). Categories: natural, suicide, homicide, accident, undetermined (report states no method/cause). Each record links to its source report.
- Caveats: counts for FY2004–FY2017 predate mandatory public reporting and vary by source; FY2025 figures are preliminary. Reconcile against primary documents before publication.
Primary source list: ICE Detainee Death Reporting · ICE ERO Statistics · Vera ICE Detention Trends · TRAC Immigration