In five days at the end of June, U.S. Immigration and Customs Enforcement arrested 10,000 people across the country. That number came from the Deportation Data Project, an effort to collect data about immigration enforcement actions across the U.S.

The project can help journalists find out how long people are held in a specific detention facility before being deported, whether someone had a criminal conviction before ICE picked them up, which field offices are driving surges in arrests and much more. It has emerged as one of the primary data sources for journalists covering the immigration enforcement.

The project seeks to make immigration enforcement data transparent and available. The team uses the Freedom of Information Act to obtain data sets directly from the government, gathers data released through others’ FOIA requests and includes data sets the government has already posted publicly.

The team behind the project

Graeme Blair, a UCLA political science professor, and David Hausman, a University of California, Berkeley law professor and former lawyer at the American Civil Liberties Union, lead the Deportation Data Project. Hausman spent years filing FOIA requests to pry immigration data out of the federal government one case at a time. In March 2025, he turned that work into something bigger.

The project is the first centralized repository of individual-level federal immigration enforcement data: every arrest, detention and deportation are logged as its own row rather than folded into a summary statistic.

Getting started

The Deportation Data Project’s webinar video on YouTube

Journalists should start by visiting the ICE data frequently asked questions page, which has explainers and a webinar on the data.

This Github repository by Raphael Romero Ruiz gives step-by-step instructions on how to navigate this data set with code. 

Data section

The data section includes both raw original data and processed, compiled data designed to facilitate analysis. The original data set is posted as the organization received it. The processed data is meant to be easier to use than the original data, by flagging possible duplicates and making dates easier to use. It is divided between Immigration and Customs Enforcement (ICE) and the Executive Office for Immigration Review (EOIR), which runs the immigration courts.

The site also contains a tools section that lets users filter, analyze and export the data.

The ICE data is divided into four categories:

Each data set’s variables are listed alongside it, and the data can be downloaded as Excel, Stata, SPSS, Parquet, R or Python files.

The original data section is divided into five categories:

Each category lists data sets by topic, with start and end dates, and a record count. Data downloads as a zip file. Throughout the processed and original data sections, the site flags any concerns or changes to the data.

Documentation section

The documentation section opens with a data guide that provides an overview of the project’s work. The project focuses heavily on individual-level data, in which each row represents a person or action. This section also includes codebooks for use with the data.

The documentation section is split into six categories:

For each of these categories, the website also offers a documents section, where users can search by document type, date or title:

Filter your jurisdiction

• • Three columns could be used during this process: Area of Responsibility, Landmark and State

• • The Area of Responsibility is the coverage area of the local ICE office, but the website warns the area can sometimes be too large 

• • The State is where the arrest occurred but may not be included in the data

• • Landmark is the place nearest to the arrest but can be inconsistently used, the website said

Limitations

• • This data is from FOIA requests, so it is not necessarily complete

• • The website warns about the reliability of the encounters and removals data set because there is conflicting information on whether the October or July data release was the most accurate

• • The new data set also does not include hospital stints because ICE omitted that data

• • ICE has also redacted some case_category values along with detention release reasons and disposition values; the site has offered solutions to fixing these issues

• • These data sets only include ICE data but not CBP, which makes up the majority of border encounters

• • apprehension_type only started being tracked in August 2025, which means it can’t be used in a multiyear analysis 

• • More caveats to the data can be found here

Story examples

Many news organizations are using this data to break stories. The New York Times built an interactive tracking individual movement from arrest to detention to deportation. The Prison Policy Initiative used it to measure how state and local cooperation with ICE changes arrest patterns. ABC7 Bay Area used it to catch a five-day surge before anyone else did.

Inside the Deportation Machine
The New York Times, Dec. 22, 2025
This interactive data-visualization piece tracks how people moved from arrest to detention to deportation, according to the Times’ methodology statement. The Times cleaned the data and removed duplicates before analyzing it.

New ICE arrest data show the power of state and local governments to curtail mass deportations
Prison Policy Initiative, Dec. 11, 2025
This brief uses Deportation Data Project data to compare ICE arrest patterns in states that cooperate with ICE against those that don’t.

ICE arrests 10,000 people across US in 5-day span, a sharp surge in Trump’s deportation push
ABC7 Bay Area, July 16, 2026
ABC7 found ICE arrested 10,000 people in five days at the end of June, and used the project’s data to measure how much deportations have increased overall.