Government Says ICE Targets Criminals, Decade of Data Says Otherwise

When the federal government defends its aggressive immigration crackdown, public safety is the go-to argument. A sweeping new analysis of ICE arrest records spanning more than a decade tells a very different story.

A study published in the Proceedings of the National Academy of Sciences examined 1.6 million Immigration and Customs Enforcement arrests recorded in administrative data from October 2015 through March 10, 2026, and found that the share of people arrested who had a criminal conviction has plummeted, even as total arrests have surged to record highs. During President Donald Trump’s second administration, ICE averaged 952 arrests per day, and with the criminal conviction rate among those arrested falling to just 34%, the majority of those arrested had no criminal conviction on their record at the time of their arrest.

As ICE has ramped up its operations to the highest levels in recent history, the agency has increasingly arrested people who have never been convicted of a crime, a pattern researchers described as “a fundamental mismatch between the justifications for intensified immigration enforcement” and the reality of who is being detained.

How Researchers Tracked ICE Arrests Over a Decade

Researchers from the University of Colorado Boulder and the University of California, Berkeley analyzed administrative data from the Deportation Data Project, which compiled records on all ICE enforcement arrests from October 1, 2015, to March 10, 2026. That amounts to 1,565,651 individual arrest records covering more than a decade of immigration enforcement across multiple administrations.

To make meaningful comparisons, the team focused on two parallel windows of time: the 414 days before and after Trump’s first inauguration, and the 414 days before and after his second. This approach gave researchers a consistent, apples-to-apples look at how enforcement changed when each administration took over. For Term 1, that window included 294,717 arrests; for Term 2, it included 520,805 arrests.

Each arrest record included information about where the arrest took place, what method ICE used, and whether the person arrested had any criminal convictions. ICE uses two main arrest approaches: arrests made when ICE takes someone directly from another law enforcement agency’s custody, and arrests where ICE agents go out into communities and make arrests on their own. Researchers tracked how each method’s use changed over time and how those shifts affected the overall conviction rate among those arrested.

Inside the ICE Arrest Surge

Before Trump’s first term, ICE averaged 304 arrests per day. During the first 14 months of that term, arrests rose to 436 per day, a 43% increase. That same baseline of 304 daily arrests preceded Trump’s second term, but this time the daily average more than tripled between January 2025 and March 10, 2026, a 214% increase.

As arrests climbed, the criminal conviction rate among those arrested dropped sharply, sinking near its lowest point across the entire decade-plus dataset.

Researchers separated two possible explanations for this decline: changes in the type of arrest tactic used versus changes in who is being targeted within each tactic. During Term 1, the conviction rate among those arrested fell by 14.2 percentage points, and almost all of that drop was due to changes in targeting, not tactics. During Term 2, the conviction rate fell by 20.6 percentage points. Again, the dominant driver was a change in who ICE is going after, with roughly three-quarters of the decline attributable to shifts in targeting, regardless of the arrest method used.

A Major Shift in How ICE Makes Arrests

One shift stands out in the way ICE’s arrest methods changed during the second term. Arrests made directly by ICE agents out in neighborhoods, rather than through cooperation with local police, jumped from 17% of all arrests in the year before Trump’s second inauguration to 47% between January 20, 2025, and March 10, 2026. No comparable shift occurred during the first term.

This matters because neighborhood arrests have consistently shown a lower conviction rate among those arrested compared to arrests made through local law enforcement partnerships. So part of the reason the overall conviction rate dropped is that ICE leaned into a method that was already less focused on people with criminal convictions.

Even so, the conviction rate fell sharply within each arrest method. Among neighborhood arrests in Term 2, the conviction rate dropped 10 percentage points. Among arrests made through law enforcement partnerships, the rate dropped by 16 percentage points. ICE is not simply using different tools; it has changed who it targets, regardless of the tool used.

Geography Shows Where ICE Enforcement Is Hitting Hardest

Enforcement patterns also varied by region. ICE operates through 25 geographic zones across the country, and the study found wide variation in how the arrest surge played out from one zone to the next.

Several zones, including those covering El Paso, Miami, New Orleans, and San Antonio, saw sharp spikes in arrests made through law enforcement partnerships. A separate cluster of zones experienced steep surges in neighborhood arrests, including those covering Boston, Chicago, Newark, Seattle, St. Paul, and Washington, D.C., as well as all zones in California and New York. Researchers noted this geographic pattern is consistent with the administration’s stated focus on so-called “sanctuary jurisdictions,” places that limit local cooperation with federal immigration enforcement.

Despite these regional differences in method, the decline in arrests of people with criminal convictions was nearly universal across the country. Among zones with the largest drops in the conviction rate, San Diego saw a 34-percentage-point decline, Washington, D.C., saw a 28-percentage-point drop, and Philadelphia saw a 27-percentage-point drop.

Researchers concluded that three forces have combined to produce the current enforcement picture: arrests have scaled to historically high levels, the arrest method has shifted toward one that yields lower conviction rates among those arrested, and in nearly every method and region, ICE is targeting people with lower conviction rates than before. All three push in the same direction, reinforcing one another.

For the millions of noncitizens living in the United States, for policymakers, and for communities on both sides of the immigration debate, the data raises a pointed question. If the central argument for record-level immigration enforcement is that it keeps communities safe by removing people who have committed crimes, the pattern in these records runs the opposite way, with most of those arrested carrying no criminal conviction at all.

Disclaimer: This article summarizes findings from an observational study of administrative arrest records. The research documents patterns and correlations in ICE enforcement data; it does not establish that any single policy caused these changes. Figures reflect the study’s dataset, which covers October 2015 through March 10, 2026.

---