If you are detained in US immigration (ICE) custody, what is the chance you die before release?
Evidence quality 4.13/5
Eight-dimension review score against the quality rubric . Each dimension scored 1–5.
- D1 Source grounding
- 5/5
- D2 Source authority
- 5/5
- D3 Arithmetic
- 4/5
- D4 Uncertainty
- 4/5
- D5 Scope
- 4/5
- D6 Prose
- 4/5
- D7 Perception honesty
- 3/5
- D8 Caveat completeness
- 4/5
≈ As likely as
Perceived
No survey asks the public to estimate the numerical chance of dying once held in ICE custody. Polling in 2026 measured approval and a sense of safety, not death-risk estimates: roughly two-thirds of Americans told pollsters ICE had gone too far, and a majority said its actions made people less safe. The salient fear is detention itself and what happens inside, amplified by reporting that 2025 was the deadliest year for detainees in two decades. Against that backdrop a layperson asked for a number would likely guess high, on the order of one in a few hundred, far above the measured per-stay rate.
Rough estimate: ~1 in 300 per detention (likely overestimate)
Source: editorial intuition, not polled
Actual
47.5 deaths per 100,000 detainee-years (FY2025)
People held in US ICE detention, fiscal year 2025
Show derivation
Subgroup = a person held in US ICE immigration detention; horizon = one detention episode, not a lifetime. Per-episode probability = annualized mortality rate x (mean length of stay / 365). JAMA (2026) reports 47.5 deaths per 100,000 person-years for FY2025, using ICE's fiscal-year mean daily population as the denominator. Mean length of stay in 2025 ran roughly 44-52 days; using 45 days: 0.000475 x (45/365) = 0.0000586, about 1 in 17,000 per stay. Low bound uses the FY2023 calm-period rate of 13.0/100,000 x 45/365 = 0.000016 (~1 in 62,000). High bound uses the winter-skewed partial-FY2026 rate of 88.9/100,000 x 45/365 = 0.00011 (~1 in 9,100). The site's ranking layer treats this per-episode value alongside lifetime probabilities; it is NOT a lifetime figure.
Caveats: The headline is a per-detention-episode probability, not a lifetime risk; the si…
The headline is a per-detention-episode probability, not a lifetime risk; the site's ranking layer places it next to lifetime figures, so it should be read as the chance of dying during one stay (~45-day mean), not over a life. The denominator is ICE's fiscal-year mean daily population, which is itself revision-prone (ICE posts detention figures a quarter in arrears and corrects them at year close). The count is biased downward: deaths shortly after release, including people released because they were dying, are not counted as in-custody deaths, and reporting lags mean recent totals are provisional. Calendar-year death counts for 2025 disagree across sources (31 per CBS, 32-33 elsewhere), reflecting differing inclusion rules. The 88.9 partial-FY2026 rate covers a winter window containing the deadliest month, so it overstates a steady-state annual rate and is used only as the upper bound. The risk has moved roughly sevenfold between FY2023 (13.0) and partial FY2026 (88.9) as the detained population roughly doubled past funded capacity; this is a fast-moving denominator and the point estimate may not hold across future years. 8D self-score clears threshold (avg 4.1, all dimensions >=3).
Related risks
Other risks on similar themes — for exploring related fears.
Landmine or UXO injury
What are the odds of being killed or injured by a landmine or unexploded ordnance?
Student visa revocation
What are the odds an international student in the US has their visa revoked or SEVIS record terminated during their program?
Visa overstay deportation
What are the odds of being detained and deported from the US after overstaying a visa?
Listeria from deli meat
How likely is a US adult to be hospitalized or die from listeria in deli meat?
Pick challenger
Once someone is booked into US Immigration and Customs Enforcement detention, the chance of dying before release is small in absolute terms and unusual in how fast it has changed. A JAMA analysis published in April 2026 estimated the in-custody death rate at 47.5 per 100,000 person-years for fiscal 2025 and 88.9 for the partial fiscal 2026, against a low of 13.0 in fiscal 2023 and a peak of 127.7 back in fiscal 2004. The denominator is ICE’s own fiscal-year mean daily population. A typical stay in 2025 ran somewhere between forty-four and fifty-two days, so converting the annualized rate to one detention episode of about forty-five days gives roughly one death per seventeen thousand stays at the fiscal-2025 rate, with a band running from about one in sixty-two thousand at the calmer 2023 rate to about one in nine thousand at the winter-weighted partial-2026 rate.
The number matters mostly as a trend rather than a level. Calendar-year 2025 saw 31 deaths by a CBS News count of ICE records, and 32 to 33 by other tallies, the most in two decades and a match for the 2004 record. That happened while the detained population roughly doubled, surpassing 68,000 by December 2025 and topping 73,000 in January 2026, far above the system’s funded bed capacity. December 2025 was the single deadliest month, with seven deaths. The rate climbed nearly sevenfold from fiscal 2023 to partial fiscal 2026, which is why a steady headline figure is fragile: the denominator is moving, the population is sicker and more crowded, and the rate is rising.
Several things push the true figure higher than the recorded one. Deaths that occur shortly after release, including cases where a detainee was released because they were already dying, are not counted as in-custody deaths. ICE posts detention data a quarter in arrears and corrects it at fiscal-year close, so recent totals are provisional. The physicians who reviewed the data described systemic weaknesses in medical care, mental-health protection, and mortality review rather than isolated lapses, and earlier work found the large majority of detention deaths in 2017 to 2021 were preventable with adequate care. No survey asks the public to put a number on this risk; polling in 2026 measured disapproval and a sense of reduced safety, not death odds, so the perceived figure here is an inference, not a measurement.
Related tidbits
Death in immigration detention runs about 47.5 per 100,000 detainee-years, or roughly 1 in 17,000 per detention episode. People picture the per-stay odds nearer 1 in 300, far above what the records show.
Claim ledger
Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.
-
[1] JAMA (Basu S, Huynh BQ, Kiang MV, Chin ET, Andrews JR) — Mortality in US Immigration and Customs Enforcement Detention
Mortality in US Immigration and Customs Enforcement Detention- Statistic
272 deaths in ICE custody FY2004 through Jan 19, 2026; mortality rate per 100,000 person-years using ICE fiscal-year mean daily population as denominator.- Excerpt
“estimate annual mortality rates among individuals detained by ICE from fiscal year (FY) 2004 through January 19, 2026, and to summarize age and cause of death patterns over time”
- Source data from
- 2026-04-16
- Accessed
- 2026-06-13
- Calculation
- Authoritative peer-reviewed primary source for the study's existence, method (FY mean daily population denominator), and total death count. Full text is paywalled; the per-year rate numbers (47.5, 88.9, 13.0, 127.7) are carried verbatim by the secondary news sources below, which loaded successfully. Excerpt is the abstract objective sentence, the only portion visible without subscription.
- Independence
- Independent academic team (Stanford/UCSF-affiliated epidemiologists); not ICE, not an advocacy group.
-
[2] ABC News — Death rates at ICE detention facilities raise concerns about health standards: Study
Death rates at ICE detention facilities raise concerns about health standards: Study- Statistic
Death rate per 100,000 person-years: 127.7 (FY2004), 13.0 (FY2023), 47.5 (FY2025), 88.9 (partial FY2026).- Excerpt
“the death rate dropped from 127.7 per 100,000 person-years in FY 2004 to 13.0 in FY 2023, then climbed to 47.5 in FY 2025 and 88.9 in partial FY 2026.”
- Source data from
- 2026-04-16
- Accessed
- 2026-06-13
- Calculation
- Provides all four fiscal-year rates verbatim from the JAMA study. FY2025 rate of 47.5/100,000 person-years is the headline point-estimate basis. The 88.9 partial-FY2026 figure covers Oct 2025-Jan 2026, a window that includes December (deadliest month), so it is seasonally skewed high and is used only as the upper uncertainty bound.
-
[3] Davis Vanguard — Immigration Detention Deaths at 22-Year High; Doctors Blame System Failures
Immigration Detention Deaths at 22-Year High; Doctors Blame System Failures- Statistic
272 deaths FY2004 through Jan 19; 18 deaths Oct 2025-Jan 2026 at annualized 88.9 per 100,000; median age of decedents 45.- Excerpt
“18 deaths from October 2025 to January 2026, at an annualized death rate of 88.9 people per 100,000.”
- Source data from
- 2026-04-01
- Accessed
- 2026-06-13
- Calculation
- Independent confirmation of the 272 total and the 88.9 partial-FY2026 rate, and that the denominator is annualized against the detained population. Used to corroborate the JAMA figures and the upper bound.
- Independence
- Independent outlet; corroborates ABC News and AOL on the same JAMA numbers.
-
[4] CBS News — ICE custody deaths are at a 2-decade high
ICE custody deaths are at a 2-decade high- Statistic
31 ICE detainee deaths in 2025 (CBS analysis of ICE records), a two-decade high; >68,000 detained as of early February 2026.- Excerpt
“In 2025, 31 ICE detainees died, a two-decade high, according to a CBS News analysis of ICE records.”
- Source data from
- 2026-03-01
- Accessed
- 2026-06-13
- Calculation
- Calendar-year 2025 death count. Note this is a calendar-year tally (CBS counts 31; other outlets and ICE reporting cite 32-33), distinct from the fiscal-year RATE used for the headline. Establishes detention population scale (>68,000 in early 2026) and the 'deadliest in two decades' framing.