What does public data show about the criminal history of people arrested by ICE during interior enforcement operations since 2025?
Executive summary
Public data from ICE and independent aggregators shows conflicting but converging patterns: ICE’s official reporting and prior-year trends emphasize arrests of people with convictions, citing figures like 71.7% convicted or charged in FY2024, while multiple public datasets and independent analyses in 2025 find that a large and growing share of people arrested in interior operations have no criminal convictions—estimates commonly cluster around 60–72% depending on the time frame and dataset [1] [2] [3] [4] [5]. Researchers emphasize geographic and temporal variation and important definitional limits in the ICE datasets that shape those headline percentages [6] [7].
1. What the public datasets themselves report about criminal history
ICE’s own annual reporting for FY2024 states that 81,312 — 71.7% — of 113,431 noncitizens ERO arrested were convicted criminals or had pending criminal charges at the time of arrest, framing recent interior enforcement as shifted toward people with serious criminal histories [1]. By contrast, public-release arrest and detention snapshots covering 2025 and compiled by independent projects show much larger shares of people with no criminal convictions among those arrested or detained during interior operations: several analyses and news organizations report that roughly two‑thirds of people in some ICE custody samples lacked criminal convictions, and that in high‑profile city deployments more than half of those arrested had no criminal record [2] [3] [4] [5]. These differences largely reflect different reporting windows and which populations are included in the calculations [6].
2. How independent analyses and journalists read the numbers
Analysts at Cato, Reason, Fortune, The New York Times and other outlets using ICE’s public tables or leaked datasets find that a majority of many 2025 interior arrests and a plurality of current detainees lacked conviction records; Cato’s leaked data and subsequent analyses put the share without convictions in the mid‑60s, and some contemporaneous reporting flagged a 71.7% no‑conviction share among detainees as of mid‑2025 [3] [4] [2] [5]. Reporters and researchers also document that violent convictions represent a small fraction of the total: analyses cited violent‑crime shares in the low single digits of the detained population [3] [4].
3. Geographic and operational variation changes the picture
State‑ and field‑office differences are pronounced: Prison Policy’s breakdown showed that some local operations and federal deployments in cities produced arrest pools with higher shares of people without convictions, while other field offices (and ICE press releases highlighting individual arrests) emphasize removals of people convicted of violent crimes—illustrating that national averages conceal sharp local variation [7] [8]. The Deportation Data Project’s ICE tables allow analysts to split periods (e.g., Jan 20–May 20 vs. May 21–Oct 15, 2025) and show that the composition of arrests shifted markedly across those windows [9] [7].
4. Official claims, messaging and counterclaims
DHS and ICE officials publicly assert that enforcement is focused on “the worst of the worst” and often count detained people with pending charges as part of the criminal cohort, a classification critics say inflates the share labeled “criminal” [10] [2]. ICE press releases also highlight arrests of individuals convicted of serious offenses to underscore public‑safety rationales, but independent datasets and reporting show such high‑profile cases are numerically a small slice of total arrests [8] [4].
5. Data limitations that shape interpretation
Public ICE datasets cover ERO interior arrests and contain fields that make disaggregation possible, but they omit border‑patrol referrals and can have missing identifiers and variable completeness across fields and offices, which complicates longitudinal and county‑level matching and leads different analysts to different denominators and time windows [6] [7]. Independent trackers also use different cutoffs and definitions (current detainees versus arrests processed that fiscal year), which explains why reported shares of people without convictions vary between roughly one‑third in some historical measures and two‑thirds in many 2025 snapshots [1] [2] [3].
6. Bottom line
Public data do not support a single, uncontested narrative: ICE’s FY2024 accounting showed most ERO arrests were of people with convictions or pending charges, but the publicly released 2025 arrest and detention tables and independent analyses document a large and growing portion of interior arrests and detainees without criminal convictions—commonly in the 60–72% range—while violent‑crime convictions constitute a small minority; local operations and definitional choices drive much of the apparent contradiction, and dataset limitations make precise, consistent national comparisons difficult [1] [2] [3] [4] [6].