Free Tool · For PTs Deciding Where to Work
The highest-paying state for a PT is not the one that pays the most.
This maps what a physical therapist's paycheck is actually worth in all 50 states and DC, after federal tax, state tax, payroll tax, and local prices. Built from BLS and BEA public data with a pinned, open-source 2026 tax model. No signup, no email required, and the whole method is published below.
Computed: … · Next scheduled refresh: …
The index
Real pay, in national-average dollars
Take a state's PT wage, subtract federal income tax, state income tax and employee payroll tax, then divide by that state's price level. What is left is what the paycheck actually buys, stated in dollars that mean the same thing everywhere. Pick a position on the wage curve to test a lower- or higher-pay scenario; percentiles are not years of experience.
On a small screen the map shows the pattern. Tap any state for its full breakdown, or use the ranked table below for the numbers.
Five rank-based favorability tiers, from least favorable to most favorable. Exact dollars and ranks remain visible; ≈ marks states within $500 of a neighboring rank, where the ordering should be treated as a close result.
Every state, ranked
Click any column to sort. This table is the same data as the map, for anyone who would rather read numbers than colour.
| # | State | Tier | Gross wage | Total tax | Take-home | Price level | Real pay |
|---|
What the numbers say
Salary is the smallest of the three things you are choosing between
Wage, tax and cost of living all vary by state. They do not vary equally. Decomposing the spread across all 51 jurisdictions at the median wage shows how much of the difference each one is responsible for.
Share of the cross-state variance in real pay, at the median wage. Exact additive decomposition of log real pay into wage, tax and price terms; the three sum to 100% by construction.
The gap between the best and worst state is about $20,000 a year
At the median wage that is the difference between roughly $88,000 and $67,000 of real annual purchasing power, for the same job. High-cost states do pay more, but not nearly enough to close the gap: wage and price level correlate at +0.58, which means about a third of a price increase comes back as pay and the rest comes out of the therapist.
The states that move furthest between a nominal ranking and a real one are the ones worth knowing about. Massachusetts and Oregon each fall 36 places once tax and prices are applied. South Dakota climbs 35, while Iowa and Arkansas climb 34. None of that is visible in a salary listing.
Wage position changes the answer more than you would expect
The map is not the same map at every point on the wage curve. Between the 10th and 90th percentile the typical state moves 9 rank places: Louisiana moves 42, Utah 31, South Dakota 30. The spread widens too, from about $23,000 to $34,000, once Nevada is set aside. Including Nevada the figure reads $62,000, but roughly 80% of that widening is Nevada's single outlying top-decile wage cell, so the smaller number is the honest one. Rank movement is the sturdier version of this finding because it does not rest on any one cell.
A state that is a good first job is not automatically a good tenth year, and the tool defaults to the median precisely because no single number covers a career.
Methodology
How the number is built
There is no composite score here, and no weights. Salary, tax and cost of living are already denominated in dollars, so they combine by arithmetic rather than by judgement. That removes the one thing an index like this is usually argued about.
real pay = (gross wage − employee payroll tax − federal income tax − state income tax) ÷ (state price level ÷ 100)
Public data sources and assumptions
- Gross wage. BLS Occupational Employment and Wage Statistics, SOC 29-1123 Physical Therapists, May 2025 reference period, accessed through the Department of Labor's O*NET OnLine. Five percentiles per state: 10th, 25th, 50th, 75th, 90th.
- Price level. BEA Regional Price Parities, all items, 2024, where the national average is 100. Dividing income by the RPP is BEA's own method for computing real personal income by state, not something invented here.
- Tax calculation. PolicyEngine US 1.764.6, an open-source microsimulation model, calculates federal income tax, employee payroll tax and each state's 2026 income tax rules. The version is pinned so this page can be reproduced.
- Geographic consistency. State income tax is used without local income tax. Maryland's state-only tax variable is selected explicitly; New York City and other municipal taxes are not inferred from a statewide record.
- Household. Single, age 30, no dependents, wage income only. Alaska's Permanent Fund Dividend is explicitly set to zero so the model does not tax income that is absent from the gross-wage numerator.
What this does not account for
- Within-state variation is larger than most state-to-state gaps. This is the biggest limitation on the page. Upstate New York and New York City are one row here and are not one market. Read a state's figure as the state average it is, not as a quote for a specific city.
- Local income taxes are not modelled. Seventeen states permit them: Alabama, California, Colorado, Delaware, Indiana, Iowa, Kansas, Kentucky, Maryland, Michigan, Missouri, New Jersey, New York, Ohio, Oregon, Pennsylvania and West Virginia. Maryland's county tax and New York City's resident tax are the two that would move a state's position most, and both would move it downward.
- This is one household scenario, not personal tax advice. Filing status, dependents, deductions, credits and other income can change both the dollar result and the ordering between states.
- Sales and property taxes are deliberately excluded from the tax term. They are already inside the price level: the RPP goods component is built on prices that include sales tax, and its housing component is rent-based. Adding them again would double-count.
- Percentiles are not experience. The wage curve is the best public proxy available at state level, but a percentile also absorbs practice setting, hours, and where in the state someone works. Treat the curve as a range, not as a years-of-service lookup.
- No sub-specialty breakdown, on purpose. BLS does not publish one for physical therapists, and the surveys that do are member-gated. Rather than invent multipliers, this version does not offer the cut.
- Student debt is not in the model. It is large and it is real, but it is roughly constant across states, so it lowers every number rather than changing the ranking.
- Wages are May 2025, prices are 2024 and tax law is 2026. These are the latest complete public vintages available for their respective inputs. The mismatch is disclosed because each source updates on a different schedule.
Every calculation is reproducible from public data. The methodology, assumptions and limitations are documented above.