GS pay scale 2026
Methodology

58 locality pay areasPage last changed 2026-10-07

How these numbers are made

Every figure on this site can be traced to a published federal source and reproduced from it. This page says exactly how, and what stops publication when something does not add up.

Who publishes this

GS Pay Tables is published by BiLingoPlus LLC, a limited liability company registered in the State of Wyoming; the about page says who builds it and how.

It is a small independent publisher. It is not a newsroom, not a consultancy, not a government body and not connected to one. Nobody pays to appear here and there is no sponsored content of any kind. The pages carry Google’s advertising code, but no advertisement is shown at the moment. When advertisements are shown, they will sit in marked places, each labeled “Advertisement”, and they will have no influence on any figure or any wording on this site.

The reason to trust a number here is not who wrote it. It is that you can check it: the source is named and linked, the arithmetic is described, and the result is reproducible. Where you find that we are wrong, tell us — corrections are made on the page, move the date that page shows, and are listed on the corrections page.

How this site is made

The tables are generated by code. Every pay rate, rank and county list on this site is produced by a program from the machine-readable files OPM, the Bureau of Economic Analysis and the Census Bureau publish, rather than retyped. Facts about pay raises — the dates of letters and orders and the percentages they set — are entered from the documents the pay raise page links to.

The text was drafted with AI assistance. The explanations were written with the help of AI tools, and the sentences that repeat across similar pages are templates into which the code puts each page’s own numbers and names.

The checks are automated. Before anything is published, ninety-two automated checks run: the first recomputes all 8,700 pay rates from OPM’s base table and compares each with the figure OPM published, and the others read the finished pages for contradictions, missing areas, broken links, mismatched dates and wording that has had to be corrected before. If any check fails, nothing is published. Errors that get past them are corrected on the page and listed on the corrections page.

Where every number comes from

Four public datasets behind the pay tables and the price comparison, all works of the United States government and all in the public domain; the further sources of the locality, state and grade pages are described below.

Salary tables and locality percentages. U.S. Office of Personnel Management, 2026 General Schedule salary tables, taken from the machine-readable files OPM publishes rather than retyped from a PDF. The list of locality codes is read from OPM’s own index page each time rather than kept in the code: OPM published tables for 47 areas in 2016, 53 in 2019, 54 in 2021 and 58 from 2024, and a fixed list would silently miss a new one.

Locality pay area definitions. OPM, 2026. These are the counties and military installations that make up each area. 55 of the 58 areas are lists of counties, 920 counties in all. The other three are drawn without counties: Alaska and Hawaii are each a whole state, and Rest of U.S. is defined by exclusion — every duty station outside the named areas.

Price levels. U.S. Bureau of Economic Analysis Regional Price Parities for 2024, at metropolitan level for the 55 areas matched to a metropolitan area and at state level for Alaska and Hawaii, where the locality area is the whole state and the state index is therefore an exact match rather than a proxy.

ZIP code relationships. U.S. Census Bureau 2020 ZCTA to county relationship file, used only to answer "which area is this ZIP in".

Each of them, and the rules the arithmetic follows, where it is published:

Where the history, the pay gap and the jobs figures come from

Each locality page also carries three blocks of its own data, and each state page one, from four more public sources. Those pages show the figures and only the sentences that single an area out; what each source is, what it cannot tell you and how to read it is here, once.

Locality pay by year: OPM tables, 2011 to 2026

The locality percentage printed in each year’s General Schedule table for each area, read from OPM’s yearly pages: machine-readable files from 2014, the PDF tables before that, and the PDF tables also for four 2024 files OPM published empty (Chicago, Davenport, Fresno and Sacramento). Each year’s area definitions give the county lists. Used for the history from 2016 on every locality page, where the chart has the same scale on every page; the 2026 figures match the 2026 tables in all 58 areas. A figure belongs to the area as OPM drew it that year: a county added later was paid another rate before it joined, and before an area had a table of its own the figure shown, in grey, and ranked is the Rest of U.S. rate its counties were paid. In 2021 and 2026 no area’s percentage changed.

The pay gap and the formula rate: Federal Salary Council memos

Attachment 1 of each memo gives, for every area, the estimated gap between non-federal and General Schedule base pay and the locality rate the Federal Employees Pay Comparability Act of 1990 would set to cut that gap to 5 percent. OPM’s council page lists memos back to 2003; the nine loaded here are the editions for pay in 2018 to 2026, so “editions checked here” on a locality page means those nine. An area that was a Rest of U.S. research area before it became a locality pay area is in the same memos’ research-area table (Attachment 2 to 5, with the gap of each of the last three years): its gap for that edition is the latest of the three, read with three PDF engines, and those memos print no formula rate for research areas. Where an edition prints an area in both tables, as it does for the areas it recommends, the two give the same gap and the row is taken from Attachment 1. “Listed as: Proposed” marks an edition in which the area is in Attachment 1 only as a recommendation: the memo calls it proposed or pending approval, or OPM did not yet pay it as a locality pay area in that year, as with Burlington, VT and Virginia Beach, VA in the edition for pay in 2018. Reno, NV and Rochester, NY have no row before the edition for pay in 2023: the earlier memos name them only among the locations that contacted the council’s staff and give them no gap. Checked: in every edition the areas’ payrolls add up to the memo’s total and their payroll-weighted gap equals its average, and the edition for 2025 matches the President’s Pay Agent’s report for 2025 row for row. Limits: the gap is the council’s model estimate, and the estimates swing between editions, for Laredo, TX from 43.50% to 79.21%, so one edition is not a trend. Since locality pay began in 1994 the formula has been overridden every year, by Congress or by the President (the 2027 decision shows what following it would cost); the Pay Agent’s report for locality pay in 2026, of 18 December 2025, declined to recommend these rates and calls the method flawed, and it has no table by area. Where an area’s gap falls below Rest of U.S.’s, the council’s rule gives it the Rest of U.S. target instead of a formula rate of its own: Corpus Christi, TX in the editions for pay in 2023 to 2026 and Palm Bay, FL in the edition for pay in 2024 (the memo for pay in 2024 names both in a note). Each area’s table shows a dash for those editions; Corpus Christi’s own formula rate is shown for pay in 2020 to 2022, and the figure printed for it in the latest memo does not even match that edition’s Rest of U.S. target.

Federal jobs: BLS Quarterly Census of Employment and Wages, 2023 to 2025

Annual averages of federal government jobs and wages by county, assigned to areas by OPM’s 2026 county lists. BLS counts every civilian federal job by the county of the worksite: the Postal Service and other pay systems are in, uniformed military and some national security agencies are not. So this is not a count of General Schedule employees, and average pay includes overtime. The 2025 average still counts staff on paid administrative leave until 30 September 2025, so it understates the drop by the end of the year, and BLS revises its figures. 103,573 jobs that BLS reports without a county in 2025 are in no area. In Connecticut BLS reports 9 planning regions instead of counties, and each is counted in the area holding most of its towns (4 straddle two areas). Civilians at Joint Base McGuire-Dix-Lakehurst, paid at the New York rate, are counted by BLS in Burlington County, New Jersey, in the Philadelphia area. Puerto Rico and the U.S. Virgin Islands are paid the Rest of U.S. rate but are not in its count. The areas and the jobs without a county add up to the sum of the state totals within 26 jobs, from rounding of annual averages.

Who is on each grade: OPM Federal Workforce Data, May 2026

The number of General Schedule employees at each grade, their median pay and the largest occupations, on the grade pages. Limits: for 43 percent of General Schedule records the duty station is withheld, so no count by area or state is published here. OPM also withholds the pay of those records: in May 2026 550,140 of the 1,279,389 General Schedule records carry no salary, every Department of Defense record among them (373,941) and most of those at the Department of Homeland Security and the Department of Justice. A grade’s median pay is therefore the median of the full-time employees whose pay OPM publishes, from 44.6% of the grade’s full-time employees at GS-12 to 77.4% at GS-6, and it is not given where fewer than 100 full-time records carry pay (GS-1: 25, GS-2: 73). It is annualized adjusted basic pay — the base rate with locality pay or a special rate, after the pay cap, without overtime or bonuses — and, with whole departments missing, not the median of everyone on the grade. OPM has also posted files for June and July 2026, but notes that some Department of Defense (Department of War) components did not submit data for those months, so May 2026 is the latest complete snapshot used here. The national median wage the grade pages compare with — $50,980 a year, $24.51 an hour, all occupations — is from the BLS Occupational Employment and Wage Statistics, May 2025.

Military grade equivalents: DoD Instruction 1000.01

The military grade beside each General Schedule grade on the grade pages is from Table 2 of DoD Instruction 1000.01, “Identification (ID) Cards Required by the Geneva Conventions” (April 16, 2012, Change 2 of June 4, 2018), and the officer titles from its Table 1. DoD sets these equivalents for the Geneva Conventions identification cards it issues to civilians, so that a card holder taken prisoner is placed in a rank category for the monthly advances of pay and the treatment by rank that the Geneva Convention on prisoners of war provides (Enclosure 2, paragraph 4). Limits: paragraph 4.b says the equivalents give civilians no rank or authority over military personnel and are not meant for any other purpose, and nothing in them is about pay, so a General Schedule grade and the military grade in its row are not paid the same. The copy read here is the Internet Archive’s capture of 27 August 2026 of the official file; every capture since January 2021 is the same file.

Hourly, biweekly and GL rates

Hourly rates are the annual rate divided by 2,087 hours and biweekly rates that hourly rate times 80, to the nearest cent, as 5 U.S.C. 5504(b) sets; the build checks them against OPM’s hourly tables for four areas. GL rates with locality pay are the GL base rate plus the area’s percentage, rounded to the dollar and held to the ceiling, and the build checks them against OPM’s law enforcement tables for five areas, every cell.

The step increase rules

5 U.S.C. 5335 and 5 CFR 531 subparts D and E, for the step increase calculator. Before publication the build runs the calculator’s own shipped code on fifteen worked cases, and on every grade and step over a grid of dates and hours, against a reference written from the rules.

Which sentences a page carries

A locality or state page prints a sentence about one of these figures only when the figure singles the area out: one of the five highest or lowest of the areas, a tie, a direction most areas did not take, an area’s first own table, a change to its county list, or a caveat without which its own figure would mislead. Every other figure stands in its table.

The arithmetic, in the order the law sets it

Locality rate = base rate, multiplied by one plus the locality percentage, rounded to the nearest dollar, and only then checked against the statutory ceiling.

That order matters. No General Schedule rate may exceed Level IV of the Executive Schedule, $197,200 in 2026. Applying the ceiling before rounding rather than after gives different figures at the top of the schedule, which is one reason published numbers occasionally disagree between reference sites. In 2026 the ceiling binds 97 cells across 37 areas.

Two derived figures follow the same discipline. The hourly rate is the annual rate divided by 2,087 hours, rounded to the cent. The biweekly rate is the hourly rate multiplied by 80 — not the annual rate divided by 26, which gives a different and wrong answer. The overtime rate follows the title 5 rule (5 U.S.C. 5542, 5 CFR 550.113): one and a half times your hourly rate while that rate stays at or below the GS-10 step 1 rate for your area, and above it the greater of your own rate and one and a half times GS-10 step 1. That is the rate for employees exempt from the Fair Labor Standards Act and the one OPM prints; an employee the Act covers is paid under whichever rule gives more (5 CFR 551.513).

Each of those three rules was derived from the published data itself and then checked against all 8,700 published cells, with no difference in any of them.

What the purchasing-power figure is, and is not

It is the salary divided by the local price level. Nothing more elaborate than that, and nothing hidden.

Locality pay is calculated from what private employers in the same region pay for comparable work. It is not a cost-of-living adjustment. Dividing the salary by the BEA price index, where 100 is the national average, gives a figure that is comparable across the country: what the salary would have to be, at average U.S. prices, to buy the same things.

Its limits are set out once, in How locality pay works. The price index used here describes 2024, two years before the 2026 pay tables. Metropolitan boundaries do not match locality pay area boundaries exactly, so the nearest metropolitan area is used as the proxy and named. And where the comparison tool or the calculator finds two areas within one percent of each other, it says the two are the same rather than pretending to a precision the data does not have.

Rest of U.S. is left without a figure entirely. It is a residual drawn from every state at once, no single price index describes it, and putting 100 there would be an invention.

What stops publication

The build fails, and nothing ships, when any of ninety-two checks finds a problem.

The first is the strictest: every rate is recomputed independently from the base table and the percentage and compared with the figure OPM published. A single cell out by a dollar stops the build. On the 2026 tables all 8,700 match.

The rest examine the rendered HTML rather than the source, because the defects that matter are the ones that survive a correct-looking build: a page that contradicts itself about whether an area gains or loses ground; an area silently missing from a table; a page with no sentence of its own that its sibling pages do not also carry; a broken internal link; text that reaches the reader as an escape sequence instead of a character; a class with no styling behind it; a request to a third-party domain. Each check exists because that exact mistake was made here at least once.

Every check is proved by deliberately breaking the source and confirming the build goes red. A check that has never failed proves nothing.

Separately, the whole site is rebuilt from an empty directory before release, so that a build that works only on one machine is caught before it is published.

What this site deliberately does not do

Take-home pay, tax, and advice.

Every figure here is gross pay: what the job pays before anything is taken out. Take-home depends on your FERS tier, your Thrift Savings Plan contribution and its type, your health and life insurance elections, your filing status and the income tax of the state you live in — which is not always the state you work in. We would rather publish one number that is exactly right than a take-home estimate resting on half a dozen assumptions invented on your behalf.

Nor does this site cover the Federal Wage System, the Senior Executive Service, Title 38 medical positions or special rate tables. Law enforcement officers are covered in one place: their special base table for grades 3 to 10, and the 2026 special rates for covered law enforcement positions, are on the law enforcement pay page, while the locality pages and the calculator use the ordinary General Schedule table. Where one of those applies to you, the figure here is a floor rather than an answer, and every page that could mislead you says so.

And nothing here is advice. GS Pay Tables publishes public federal data and arithmetic performed on it. Questions about your own pay belong to your servicing human resources office. Questions about this site belong to hello@gspaytables.com.