Open data
Free to cite. Please attribute and link, and state the tax year — these numbers change every January.
Machine-readable
What Americans actually check about their own pay — the observed dataset, not the computed one. Every other figure on this site is derived from published tax tables and could be reproduced by anyone holding them; this one exists nowhere else. CC BY-NC-SA 4.0 — free for research, journalism and personal use, commercial use needs a licence — CORS open, and the methodology travels inside the file rather than as a link, because a link means nothing once the file has been downloaded.
It counts CALCULATIONS, not page views, so it will read far lower than any page-analytics product — that gap is the finding, not a defect. Income exists only as one of eight ranges. Thin categories are withheld and the withheld totals are published alongside, so you can see what is missing rather than assume nothing is.
Licensed data for organisations
The open file is the coarse view: totals by state, by income range, by the kind of pay people are checking. Under licence we provide extracts that no public statistic carries, because they come from what people do at the moment they check their own pay:
- Neighbourhood demand — calculations by ZIP3 area and income range, month by month.
- Relocation intent — which state pairs get compared inside one visit, and at what income.
- The expectation gap — who finds their take-home lower than they thought, by state and income.
- Work patterns — shift hours, overtime, tips and shift premiums, by state and pay band.
Always aggregates, never individual records. No extract contains a salary — income exists only as one of eight ranges — and any figure for a ZIP3 area needs at least twenty calculations behind it, with smaller cells withheld and the withheld totals published alongside.
What it is not: a random sample of Americans. It is the people who used this calculator, which skews to whoever finds a pay calculator, and every extract says so in the file. Anyone who would be embarrassed by that sentence appearing in their own write-up should not license it.
Licensees agree not to attempt re-identification and not to resell the raw extracts. Ask about a licence.
What is here
- 824 occupations
- 51 jurisdictions (50 states + DC)
- 34,524 occupation-and-state wage figures, each with tax computed
- Tax year 2026, all four filing statuses
Headline figures
Take-home on a $75,000 salary, single filer, 2026:
| Rank | State | Take-home |
|---|---|---|
| 1 | Alaska | $61,593 |
| 2 | Florida | $61,593 |
| 3 | Nevada | $61,593 |
| 4 | New Hampshire | $61,593 |
| 5 | South Dakota | $61,593 |
| 47 | Massachusetts | $57,718 |
| 48 | Hawaii | $57,696 |
| 49 | Minnesota | $57,686 |
| 50 | Delaware | $57,684 |
| 51 | Oregon | $56,088 |
Sources
- Wages: U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics (OEWS), May 2025 state estimates, May 2025 estimates — original bulk file
- Federal tax: IRS Revenue Procedure 2025-32
- Social Security wage base: Social Security Administration
- State tax: state revenue department published rate tables
The tax figures are ours — computed by our own engine, not copied from another calculator. The methodology is written up in full on the methodology page, including what it does not model.
Anonymous usage data
We also publish what people calculate, in aggregate — which states and which salary brackets get checked. That dataset is anonymous by construction, not by promise: exact salaries are bucketed in the browser and never transmitted. See privacy for exactly what is and is not recorded. Aggregates will be published here once there is enough volume for them to mean anything; publishing thin data would be worse than publishing none.