What percentage of deportations under Trump were due to non-violent crimes?

1. What the numbers in the record actually say

Nonpublic ICE data analyzed by the Cato Institute found 65% of people taken into ICE custody had no criminal convictions and that 93% had no convictions for violent offenses, a framing the group used to argue that detentions were largely non‑violent in nature ^[1]. Major news organizations and fact‑checks report similar proportions: AP and Fortune reported that, among detainees with convictions, 6.9% were for violent crimes and 53% were for nonviolent categories like immigration, traffic or vice offenses ^{[2] [3]}. Snopes summarized ICE’s June 2025 snapshot as showing 72% of people in ICE facilities had no criminal convictions, while noting the dataset mixed people detained before and during Trump’s second term ^[4].

2. How “non‑violent” is being defined and why that matters

Reporting groups separate violent convictions from nonviolent ones; nonviolent convictions often include immigration‑related offenses, traffic violations, and “vice” crimes—categories that range widely in severity and public safety implications ^{[2] [1]}. Because different analyses use different denominators (all detained vs. only those with convictions) the headline percentage for “non‑violent” can change: one result—53% nonviolent—refers only to those with convictions, while Cato’s 65% no convictions statistic covers the entire detained population ^{[1] [2]}.

3. Limits of the available data and competing readings

Several outlets caution about dataset limitations: ICE’s public counts include people detained before Trump’s second term, and agency statistics mix long‑term detainees with recent arrests, so summer‑2025 snapshots do not represent a pure Trump‑era cohort ^[4]. CBS and other reporting note fluctuating populations and that public datasets don’t always distinguish charges from convictions or capture foreign criminal records and INTERPOL notices, which DHS officials say can include people with serious offenses not reflected in ICE conviction counts ^{[5] [6]}. Snopes emphasized that the 72% figure includes people detained before the administration and that the detention population is dynamic ^[4].

4. Political narratives and the media context

The administration has framed deportations as targeting the “worst of the worst,” while watchdogs and many media outlets say the data do not support that characterization at scale ^{[6] [3]}. Pro‑administration outlets emphasize arrests of people convicted of rape, assault and other violent crimes to support the policy’s public‑safety rationale ^[7]. Advocacy groups and immigrant‑rights groups argue the enforcement sweep is broad, detaining many without convictions and resulting in family separations and other harms ^{[8] [9]}. Each side selects data points that fit its framing—DHS highlights violent‑offender arrests; independent analysts highlight the high share of detainees without violent convictions ^{[6] [1]}.

5. What this means for the core question — percentage due to non‑violent crimes

If you measure by detainees with convictions, roughly 53% of convictions reported in coverage were for nonviolent crimes while only about 6.9% were violent ^{[2] [3]}. If you measure across the entire detained population, multiple analyses put a majority with no criminal convictions at all (Cato: 65% no convictions; Snopes citing ICE: ~72% with no convictions) — which implies an even larger share were not detained for violent crimes ^{[1] [4]}. Available sources do not provide a single definitive “percentage of deportations due to non‑violent crimes” that isolates only removals (as opposed to detentions) and fully adjusts for pre‑existing detainees, foreign convictions, or subsequent legal outcomes ^{[4] [5]}.

6. Bottom line and how to read future claims

The weight of current reporting indicates most people ICE detained in mid‑2025 lacked violent criminal convictions; many had no convictions at all and among those convicted the plurality were nonviolent offenses ^{[1] [2] [3]}. However, datasets are heterogeneous and politically contested: check whether a claim refers to “detainees” versus “deportations/removals,” whether it counts convictions or charges, and whether it isolates the administration’s period — because those choices materially change the percentage you will see ^{[4] [5]}.