The federal workforce shrank by 10.3% in 2025, or a net of nearly 238,000 workers, according to a Pew Research Center analysis of recently published government data.
| Year | Federal workers | New hires | Separations |
|---|---|---|---|
| 2016 | 2093868 | ||
| 2017 | 2081160 | ||
| 2018 | 2097881 | ||
| 2019 | 2131465 | ||
| 2020 | 2171513 | ||
| 2021 | 2170157 | ||
| 2022 | 2176983 | ||
| 2023 | 2261282 | ||
| 2024 | 2312301 | 263433 | -192562 |
| 2025 | 2074649 | 116912 | -348219 |
A total of 348,219 people quit, retired, were laid off or otherwise left federal employment last year – an 80.8% increase from 2024. At the same time, 116,912 people started working for the federal government – a 55.6% decrease from the year before.
The job losses affected a broad swath of federal workers, with few noteworthy distinctions among people of different educational attainment, length of federal service and type of appointment (that is, the competitive civil service or otherwise).
However, the Trump administration’s job cuts disproportionately affected younger and less experienced federal workers.
People younger than 35 made up 18% of the federal workforce at the end of 2024, but that figure had slipped to 16.8% by the end of 2025. And workers with less than two years of experience – most of whom would still have been on probation and without full civil service protections – fell from 16.2% of the federal workforce to 10.3%.
The Office of Personnel Management, the federal government’s human resources department, no longer publishes data on the gender, race, ethnicity or disability status of federal workers. This information was available when we performed a similar analysis in January 2025.
Which federal agencies saw the biggest decreases in workers?
Among major agencies, the Education Department and the U.S. Agency for International Development (USAID) – two of President Donald Trump’s earliest and most frequent targets – had the steepest job cuts.
| Department/Agency | Group | Dec 2024 | Dec 2025 | % change |
|---|---|---|---|---|
| Veterans Affairs | Dept | 480075 | 451121 | -6.0% |
| Homeland Security | Dept | 231337 | 227584 | -1.6% |
| Army* | Dept | 222831 | 198448 | -10.9% |
| Navy* | Dept | 221918 | 205643 | -7.3% |
| Air Force* | Dept | 173456 | 156678 | -9.7% |
| Defense | Dept | 159983 | 146609 | -8.4% |
| Justice | Dept | 117379 | 107415 | -8.5% |
| Treasury | Dept | 117063 | 89881 | -23.2% |
| Health and Human Services | Dept | 93035 | 75134 | -19.2% |
| Agriculture | Dept | 91047 | 72049 | -20.9% |
| Interior | Dept | 64082 | 56872 | -11.3% |
| Transportation | Dept | 57462 | 53512 | -6.9% |
| Commerce | Dept | 48804 | 42084 | -13.8% |
| Energy | Dept | 17607 | 14955 | -15.1% |
| Labor | Dept | 14504 | 12421 | -14.4% |
| State | Dept | 14485 | 11713 | -19.1% |
| Housing and Urban Development | Dept | 8842 | 6299 | -28.8% |
| Education | Dept | 4273 | 2453 | -42.6% |
| Social Security Administration | Agencies | 57952 | 50718 | -12.5% |
| National Aeronautics and Space Administration | Agencies | 17999 | 16869 | -6.3% |
| Environmental Protection Agency | Agencies | 16990 | 14661 | -13.7% |
| General Services Administration | Agencies | 13391 | 10346 | -22.7% |
| Small Business Administration | Agencies | 8611 | 5779 | -32.9% |
| Federal Deposit Insurance Corporation | Agencies | 6559 | 5626 | -14.2% |
| Smithsonian Institution | Agencies | 4489 | 4228 | -5.8% |
| Securities and Exchange Commission | Agencies | 4866 | 3992 | -18.0% |
| Consumer Financial Protection Bureau | Agencies | 1752 | 1248 | -28.8 |
| Peace Corps | Agencies | 985 | 709 | -28.0 |
| U.S. Agency for International Development | Agencies | 4895 | 370 | -92.4% |
The Education Department’s staff shrank 42.6% between December 2024 and December 2025, from nearly 4,300 employees to fewer than 2,500. And USAID was all but eliminated, going from just under 4,900 employees to 370 (-92.4%).
Related: What the data says about the U.S. Department of Education | What the data says about U.S. foreign aid
Other deep cuts came at the parent agency of the National Endowments for the Arts and the Humanities (-56.6%), AmeriCorps (-43.6%), the Small Business Administration (-32.9%), the agency that oversees Voice of America and other international broadcasters (-32.7%), and the National Science Foundation (-30.3%).
Several of the Office of Personal Management’s data categories are heavily redacted, limiting our analysis. For instance, nearly half of the December 2025 records don’t list workers’ job locations or pay.
And while the dataset covers most of the executive branch, some agencies are omitted entirely. Those include various intelligence agencies and most of the Executive Office of the President (with the notable exception of the Office of Management and Budget). The dataset doesn’t include Congress or any legislative agencies, the federal judiciary, or such quasi-independent entities as the U.S. Postal Service and the Federal Reserve.
The dataset does include the Consumer Financial Protection Bureau (CFPB), an independent agency that organizationally is part of the Federal Reserve. The CFPB’s headcount fell 28.8% last year.
Though overall headcount at the Department of Homeland Security fell slightly, that wasn’t the case at its immigration-enforcement components. Immigration and Customs Enforcement added about 7,500 workers, ending 2025 at 28,272 – a 36.1% increase – and Customs and Border Protection rose from 66,613 employees to 67,587 (up 1.5%). But most other Homeland Security components lost jobs, including the Transportation Security Administration (-4.3%), Citizenship and Immigration Services (-11.4%), and the Federal Emergency Management Agency (-14.0%).
Which occupations saw the biggest decreases?
The job cuts affected more white-collar workers, who made up 92.1% of the federal workforce at the end of 2024, than blue-collar workers. Headcount in white-collar occupations fell 10.6% last year, versus 6.7% among blue-collar occupations.
Among the larger occupational groups, some of the largest reductions were in:
- Information and arts: -20.6%
- Accounting and budget: -17.3%
- General administrative, clerical and office: -17.0%
- Business and industry: -16.4%
- Human resources management: -15.3%
One of the few white-collar occupational groups to remain stable was the “investigation” group, due largely to increases in border patrol enforcement (up 797 employees), customs and border protection (+570), and related fields, such as “general inspection, investigation, enforcement, and compliance” (+2,035). Those more than offset decreases in fields such as compliance inspection and support (down 2,177 employees), food inspection (-349), aviation safety (-141) and equal opportunity investigation (-122).
Other notable losses in specific occupations included:
- Medical workers, including nurses (-4,038), medical officers (-2,318) and practical nurses (-1,046)
- Tax professionals, including internal revenue agents (-3,141), internal revenue officers (-1,030) and tax examiners (-4,309)
- Information technology managers (-13,986)
- General attorneys (-6,608)
- Contact representatives, who explain policies and provide benefit administration to the general public (-6,629)
Facts Only
The federal workforce decreased by 10.3% in 2025, resulting in a net loss of nearly 238,000 workers.
In 2025, 348,219 federal employees left their positions, an 80.8% increase from 2024.
New hires in 2025 totaled 116,912, a 55.6% decrease from the previous year.
Workers under 35 made up 18% of the federal workforce in 2024 but dropped to 16.8% by 2025.
Employees with less than two years of experience fell from 16.2% to 10.3% of the workforce.
The Education Department's workforce shrank by 42.6%, from 4,273 to 2,453 employees.
USAID's workforce decreased by 92.4%, from 4,895 to 370 employees.
The Treasury Department saw a 23.2% reduction, while Health and Human Services lost 19.2% of its staff.
Immigration and Customs Enforcement added 7,500 workers, a 36.1% increase.
White-collar occupations experienced a 10.6% decline, compared to 6.7% for blue-collar roles.
Occupations like information and arts (-20.6%), accounting and budget (-17.3%), and human resources management (-15.3%) saw significant reductions.
The Office of Personnel Management no longer publishes demographic data on federal workers.
Executive Summary
The federal workforce experienced a significant reduction in 2025, declining by 10.3% or nearly 238,000 workers, according to government data. The cuts were broad-based but disproportionately affected younger and less experienced employees, with workers under 35 and those with less than two years of service seeing notable declines in representation. Agencies like the Education Department and USAID faced the steepest reductions, with USAID's workforce shrinking by 92.4%. While some areas, such as immigration enforcement within the Department of Homeland Security, saw increases, most agencies experienced job losses, particularly in white-collar occupations like accounting, human resources, and information technology. The data also highlights gaps in reporting, as demographic details on gender, race, and disability status are no longer publicly available, limiting deeper analysis of the cuts' impact.
The reductions reflect a shift in federal priorities, with certain enforcement and security roles expanding while administrative, educational, and regulatory functions contracted sharply. The lack of transparency in some data categories, including job locations and pay, further complicates efforts to fully assess the changes. The overall trend suggests a strategic realignment of the federal workforce, though the long-term implications for government operations and public services remain unclear.
Full Take
The strongest version of this narrative presents a clear and data-driven account of a significant reduction in the federal workforce, with specific emphasis on the disproportionate impact on younger and less experienced workers. The analysis highlights agency-specific cuts, particularly in education and international aid, while noting exceptions in immigration enforcement. The inclusion of raw numbers and percentages strengthens the credibility of the report, and the acknowledgment of missing demographic data adds a layer of transparency about the limitations of the analysis.
However, the narrative could be vulnerable to patterns of distortion if the focus on specific agencies (e.g., Education, USAID) is framed to imply a targeted political agenda without explicit evidence of intent. The absence of demographic data might also be exploited to obscure broader societal impacts, though the article itself does not engage in such manipulation. The emphasis on enforcement growth within Homeland Security, juxtaposed with cuts elsewhere, could be interpreted as a forced binary—suggesting a trade-off between security and other government functions—though the article presents this as factual rather than normative.
Root causes likely include policy shifts prioritizing certain federal functions over others, with historical echoes of past administrations' efforts to reshape the bureaucracy. The implications for human agency are significant: younger workers and those in probationary roles face heightened job insecurity, while the public may experience reduced access to services in areas like education and healthcare. The beneficiaries of these changes are unclear, though the expansion of immigration enforcement suggests a prioritization of border security.
Bridge questions: What policy goals might explain the uneven distribution of cuts across agencies? How might the lack of demographic data affect our understanding of equity in these workforce changes? What would it mean if these trends continue over the next decade?
Counterstrike scan: A coordinated influence campaign might amplify the cuts to education and aid while downplaying enforcement growth, framing the changes as ideologically driven. However, the article avoids overt partisan framing and presents data neutrally, with no clear alignment to such a playbook.
Patterns detected: none
Sentinel — Human
The article exhibits strong hallmarks of human authorship, including data-driven specificity, contextual nuance, and stylistic variability. No significant synthetic signals detected.