Best-Paying State for Every US Occupation

This dataset names the best-paying state for each of 824 US occupations, measured by the annual 90th-percentile wage in the Bureau of Labor Statistics Occupational Employment and Wage Statistics release of May 2025. It differs from the raw BLS answer in one respect: a state must employ at least 500 people in an occupation before it can be named.

What a 500-person floor changes

Ask the May 2025 OEWS release which state pays door-to-door sales workers, news and street vendors the most, and the answer is Missouri, at $113,810 a year for the 90th percentile of the wage distribution. Ask the same release how many door-to-door sales workers Missouri employs, and it does not publish a number: the estimate rounds to zero. Both facts come from the same release.

That is not a BLS error. OEWS publishes a state wage estimate wherever the survey collected enough responses to compute one, and it makes no claim that the state is a meaningful part of the occupation. A wage built from a handful of returns is printed on the same footing as one built from thousands. For occupations with hundreds of thousands of workers, this rarely matters. For the long tail it matters a great deal, because the state that tops the wage table is frequently the state with the thinnest sample.

PayCrunch requires a state to employ at least 500 people in an occupation before it is quoted as the best-paying state. Applied across all 824 occupations in the release, that floor changes which state wins for 389 of them. In every one of those 389 cases the state the unfiltered data picks employs fewer than 500 people in the job; in 43 of them BLS publishes no employment estimate for that state at all, and in a further 89 the estimate is under 100 people. The median gap between the unfiltered figure and the filtered one is $18,380 a year. The largest is $376,450, for dentists in specialties other than the named ones, where the unfiltered answer is $784,410 in New Jersey — from 290 people — and the filtered answer is $407,960 in Texas.

Some of the individual cases:

None of this is a discovery. It is a filter applied to public data, and the filtered number is still a survey estimate with all the uncertainty that implies. What the filter buys is narrower: when a PayCrunch page says an occupation pays best in a particular state, that state is one where the occupation demonstrably exists.

Method

Source

U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025 release — the national and state estimate files. Nothing here is a PayCrunch estimate; every wage figure in the dataset appears in that release. The derived work is the selection rule.

The measure

Ranking uses the annual 90th-percentile wage. That is the wage level below which 90 percent of the estimated wage distribution for that occupation in that state falls. It is a point on a distribution, not a ceiling and not what anyone in particular earns. It is used here rather than the median because the question the dataset answers is where the pay range in an occupation reaches highest, and because the median is also published in every row for anyone who wants to rank on that instead.

The floor, and why 500

OEWS publishes state wage estimates without regard to how large the occupation is in that state. Ranking on those estimates unfiltered produces winners drawn from very small samples, which is what the 389 changed answers above are made of. The rule applied here is that a state qualifies only if OEWS reports employment of 500 or more in that occupation in that state; among qualifying states, the one with the highest annual 90th-percentile wage is named.

500 is a judgement, not a standard, and it is a trade. It keeps 769 of the 824 occupations on a named state while removing the estimates most likely to be artefacts of a thin sample; the remaining 55 lose their state answer entirely. A lower floor would keep more occupations on a named state and readmit some of the estimates the filter exists to remove. A higher floor would remove more artefacts and push more occupations to the national fallback. The published file cannot show you the counts at other settings, because it records only the winner at this one — but every row carries the unfiltered winner alongside the filtered one, so the effect of the floor is visible for each occupation, and anyone who prefers the unfiltered answer can take it straight from the file.

When no state qualifies

For 55 occupations, no state reports 500 or more people employed. Those rows fall back to the national 90th-percentile wage, best_state is left empty and scope is set to national. Commercial divers are one: the unfiltered answer is $250,960 in Washington, a state with no published employment estimate for the occupation, and the fallback is the national figure of $157,790. Across the 769 rows that do resolve to a state, 48 different states are named, counting the District of Columbia.

The full table

All 824 occupations. Click a column heading to sort; type to filter by occupation title. The floor marker means the 500-employee floor changed which state is named; national means no state qualified.

Rows are read at page load from /data/open/best-paying-states.json. Employment shown as "n/a" is employment BLS does not publish for that state and occupation.

Download

Both files carry the same 824 records and the same fields. The JSON opens with a metadata header describing the source, the version, the employment floor and every field.

License: CC BY 4.0. Use it commercially, redistribute it, build on it. The one condition is attribution: please credit paycrunch.co and link back to this page. The underlying wage and employment estimates are public-domain US government data from the Bureau of Labor Statistics; what is licensed here is the derived selection.

How to cite

Ready to paste:

PayCrunch (2026). Best-Paying State by Occupation, United States, version 2026-08-26. Derived from U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025. https://paycrunch.co/best-paying-states.html (accessed 26 August 2026).

If you quote a single figure, cite the occupation and note the floor — for example: "Texas, at $104,420 for the 90th percentile, among states employing at least 500 geological technicians (PayCrunch 2026, from BLS OEWS May 2025)." Please cite BLS as well as PayCrunch; the wage figures are theirs.

Limitations

• OEWS is a survey. Every figure carries sampling error, and the error is largest exactly where the occupation is smallest. The floor reduces how often that error decides the answer; it does not remove it. BLS publishes relative standard errors alongside the estimates, and they are worth consulting for any single figure that matters to you.
• The 90th percentile is not a maximum. It is a point on a wage distribution. Pay above it exists, and no one should read it as what a job "pays".
• No cost-of-living adjustment. State figures are nominal. A wage in California and the same wage in Iowa are not the same standard of living, and this dataset makes no attempt to say so.
• 500 is a judgement call, not a standard. BLS does not endorse it and no statistical convention requires it. It was chosen because it removes most of the visibly thin estimates while leaving the large majority of occupations on a named state. A different analyst would reasonably choose differently, which is why every row also carries the unfiltered answer.
• 55 occupations have no state answer. They fall back to the national 90th percentile. Those rows should not be read as "the whole country is the best-paying state" — they mean the floor left nothing to name.
• Occupation titles are SOC titles. They are classification labels, not job titles as employers use them, and one SOC code can cover several distinct jobs.
• One release only. These are May 2025 figures. Comparing them with an earlier OEWS release is not straightforward: SOC definitions and estimation methods change between releases.

Occupations in this dataset

Individual PayCrunch pages for some of the occupations discussed above: