The federal count dropped by roughly 16,000
The average number of noncitizens incarcerated by the Federal Bureau of Prisons fell from about 36,300 in fiscal 2017 to 20,300 in fiscal 2024, a 44% decline. More than half of their convictions were for immigration offenses and 30% were drug-related.
ICE removed about 84,800 of the 111,200 noncitizens who completed at least one federal prison term during fiscal 2017–24, as measured in December 2025—76%. The other 26,400 should not all be labeled unlawfully present: GAO’s noncitizen category includes lawful status, and removal can depend on case and country conditions.
| Population or cost | Earlier | Later / total |
|---|---|---|
| BOP noncitizen average | 36,300 (FY17) | 20,300 (FY24) |
| SCAAP-eligible incarcerations | 128,000 (state FY16) | 73,500 (state FY22) |
| ICE removals after BOP term | — | 84,800 of 111,200 |
| DOJ spending | — | More than $9B, FY16–23 |
The state-local measure covers only a slice
GAO found no reliable comprehensive citizenship data for everyone incarcerated by states and localities. It used the State Criminal Alien Assistance Program, which covers only incarcerations that meet reimbursement criteria. That count fell 43%, from about 128,000 in state fiscal 2016 to 73,500 in 2022.
Because SCAAP is partial, it cannot be added to the federal average to produce a national total or used to infer a rate for all state prisoners.
$9 billion excludes unreimbursed local cost
DOJ spent more than $9 billion from fiscal 2016 through 2023: roughly $8 billion for BOP incarceration and $1.44 billion in SCAAP reimbursements. GAO cautioned that the figure excludes costs states and localities paid without federal reimbursement.
The finding therefore supports a federal spending floor, not an all-government cost estimate. It also does not measure costs or savings from detention before conviction or immigration custody after prison.
Method and limitation
The newsroom compared the two incarceration series without merging them, calculated the difference between completed terms and recorded removals, and checked GAO’s definition of noncitizen. GAO used the most recent available datasets, whose end years differ.
Counts of incarcerations are not counts of unique people in every dataset, and conviction categories do not establish immigration status.
Sources and further reading
GAO: Noncitizen incarcerations, convictions, removals and costs ↗