Immigration & Crime in the United States 2010–2025
A primary-source audit of what the public data actually shows about immigration status and crime. Federal sentencing, federal prison, ICE enforcement, and the one state — Texas — that comprehensively tracks offender immigration status. Plus a side-by-side panel testing the loudest assertions from both directions against the documented record — with each source's institutional priors disclosed.
State-by-state · how each state tracks offender immigration status
Hover a state for detailsWhat the primary data shows
The numbers below come from the most recent published editions of each source as of May 2026. Each row is sourced and dated. Where the data has known limits, the methodology section explains them in plain English.
Federal Sentencing — USSC FY2025 Sourcebook (published Apr 2026)
Federal Bureau of Prisons — Citizenship (per GAO-24-107598)
ICE ERO — FY2024 + preliminary FY2025
Texas DPS — June 2011 to Feb 2025
Per-100,000 rate comparisons — Texas
Two of the most-cited rate analyses of the Texas data, both using DPS source data:
| Source / Year | Native-born | Legal immigrants | Illegal immigrants |
|---|---|---|---|
| Cato Institute (libertarian) — total criminal conviction rate, conviction year 2018 (per 100K) | 1,422 | 535 | 782 |
| Cato Institute (libertarian) — homicide conviction rate, 2013–2022 (per 100K) | 3.0 | 1.2 | 2.2 |
| Light, He & Robey (PNAS 2020, peer-reviewed) — violent crime arrests, 2012–2018 | Native-born 2× more likely to be arrested for violent crime, 2.5× for drugs, 4× for property crime than undocumented. | ||
The Center for Immigration Studies — a restrictionist policy organization — argues this data understates illegal-immigrant criminality, because Texas reclassifies offenders from "other/unknown" to "illegal" over time and many true illegal aliens sit in the "other" bucket at any given moment. The Cato Institute's reply (Nowrasteh) — a libertarian, generally pro-immigration policy organization — documents two specific errors in CIS's recalculation: (a) double-counting of some homicide defendants and (b) use of a population denominator inconsistent with CIS's own estimates. Both organizations bring known priors. Both critiques have technical merit. Neither overturns the underlying pattern that Texas's per-capita conviction rates for illegal immigrants are lower than for native-born citizens, and the reader should weigh both organizations' framing accordingly. The peer-reviewed PNAS analysis above does not depend on either think tank's reanalysis.
A note on Indiana
I write from Indiana. The Indiana Department of Correction does not publicly track or publish offender immigration status. There is no Indiana equivalent of the Texas DPS Criminal Illegal Noncitizen Data portal. That means Hoosiers asking "what does the data say about illegal-immigrant crime in our state?" do not currently have a primary-source answer to turn to. They get federal aggregates, Texas extrapolations, and partisan claims — none of which describe Indiana. The 49-state data gap is not somebody else's problem. It is also Indiana's.
Claims vs the Documented Record
Eight of the most-repeated claims, tested against primary federal and state data. Selected to span both directions of the debate — not to argue one side. Each card shows the claim, the closest thing to a "documented record" available from primary sources, and the gap between the two. Where think-tank analyses are cited, their institutional priors are disclosed.
Reading the Cato / CIS dispute in plain English
The two largest think-tank analyses of Texas data disagree on one specific calculation. Here it is in the simplest possible terms. Both organizations have institutional priors: Cato is libertarian and generally pro-immigration; CIS is restrictionist. Both critiques have technical merit. The specific factual claims below about CIS's errors are documented in Alex Nowrasteh's (Cato) reply to CIS; a careful reader should follow the link and judge.
What Texas DPS records
When someone is arrested in Texas, DPS pings DHS IDENT (the federal fingerprint database) to find out their immigration status. If DHS has a record, the arrestee gets tagged with the correct status. If DHS has no record (because the person was never previously fingerprinted by an immigration encounter), the arrestee is logged as "other / unknown" and may later be reclassified if their status becomes known.
Where the two sides disagree
CIS argues: at any given snapshot, the "other/unknown" bucket contains many true illegal immigrants who have not yet been reclassified. So the published illegal-immigrant numerator is too low.
Cato (Nowrasteh) responds: that's partly true, but when CIS recalculated the homicide conviction rate to "fix" this, they (a) double-counted some defendants and (b) used a denominator (Center for Migration Studies population estimate) that conflicts with CIS's own population numbers used elsewhere.
What's left when the dust settles
Both sides agree the Texas data has measurement limitations. Both agree the data does not show illegal immigrants committing crime at higher per-capita rates than native-born citizens. They disagree on how much lower the rate actually is. That gap matters, but it does not change the direction of the finding. The peer-reviewed Light, He & Robey analysis — which does not depend on either think tank — points the same direction.
Methodology
What this page is
An audit of publicly available primary sources on immigration status and crime in the United States. Federal data (USSC, BOP, BJS, ICE ERO) covers federal arrests, sentencings, and incarcerations. State data is dominated by Texas because Texas is effectively the only state with a public, longitudinal, comprehensive offender-status dataset.
What this page is not
It is not a national rate calculation. It is not a single causal study. It is not a defense or indictment of any specific policy. Anyone who reads it that way is reading past what the page says.
Source-selection rules
- Federal government datasets are preferred over derivative analyses.
- Peer-reviewed studies are cited alongside any think-tank reanalysis.
- Where a think tank or advocacy organization is cited, its institutional priors are disclosed at first mention (Cato: libertarian; CIS: restrictionist; CAP: progressive; MPI: non-partisan immigration research).
- No single-incident news story is treated as evidence of a rate.
- No social-media-sourced claim is taken without a primary citation.
Known limits, stated up front
- Texas DPS data only counts persons previously in the DHS IDENT database. Illegal entrants without a prior DHS encounter who commit a state crime are not counted as illegal — they appear as "other/unknown."
- Status reclassification is lagged; the "other/unknown" bucket likely contains some true illegal immigrants at any given moment.
- Unauthorized-population denominators vary across credible sources by 1–2 million people, which changes rate calculations.
- Forty-nine states do not publish offender immigration status in a usable form, so national rate claims rest on extrapolation. GAO has independently confirmed that federal agencies do not collect citizenship information for the roughly 90% of incarcerated people held in state and local facilities (GAO-24-107598).
A closing note on evidence and certainty
The data is not as clean as anyone wants it to be. One state — Texas — does the work of tracking offender immigration status publicly and comprehensively. The federal government tracks its own slice. Forty-nine other states do not publish this data in a usable form, and that absence is itself a fact worth sitting with.
What the data we do have shows is this: in Texas, across multiple independent analyses using the same source — including one peer-reviewed PNAS study and two think-tank analyses with opposing institutional priors — illegal immigrants are convicted of crimes at lower per-capita rates than native-born citizens. The Center for Immigration Studies (restrictionist) disputes the calculation. The Cato Institute (libertarian) disputes the dispute. Both critiques have technical merit; neither overturns the underlying pattern.
The strongest honest claim is not "illegal immigrants commit more crime" or "illegal immigrants commit less crime." It is: the data we have, with all its limits, does not support the higher-rates claim, and the absence of better data in 49 other states is a policy choice we should examine.
I am open to being shown otherwise — with primary sources. That is the standing offer for this page.
References & further reading
Federal data (primary)
- USSC 2025 Sourcebook of Federal Sentencing Statistics (published April 2026, covers FY2025)
- USSC: Federally Sentenced Non-U.S. Citizens
- BOP Inmate Citizenship Statistics
- BJS Federal Prisoner Statistics under the First Step Act, 2025
- ICE Enforcement and Removal Operations Statistics
- DHS Office of Homeland Security Statistics — Monthly Enforcement Tables
- GAO-24-107598: Noncitizens in the U.S.: Public Information on Federal Incarcerations (September 2024)
State data (primary)
Peer-reviewed studies
- Light, He & Robey 2020 — Comparing crime rates between undocumented immigrants, legal immigrants, and native-born US citizens in Texas (PNAS)
- Sanctuary policies reduce deportations without increasing crime (PNAS 2020)
- Local effect of sanctuary policies on crime (Applied Economics Letters, 2025)
- UNM 2024 — No correlation between sanctuary cities and crime
Think-tank and advocacy analyses (priors disclosed)
- Cato Institute (libertarian / pro-immigration): Illegal Immigrant Incarceration Rates, 2010–2024
- Cato Institute (libertarian): Criminal Immigrants in Texas
- Center for Immigration Studies (restrictionist): Misuse of Texas Data Understates Illegal Immigrant Criminality
- Alex Nowrasteh (Cato, libertarian) reply to CIS
- Center for American Progress (progressive): 2017 county-level detainer-cooperation crime analysis