Methodology

What Physician Salary Numbers Actually Measure

Medscape, Doximity, MGMA, AMGA and SullivanCotter report different physician pay figures for the same specialty. Here is what each one counts, and what none of them can tell you.

Person counting a stack of dollar bills at a table

Figures as of October 2026.

Medscape's 2026 report puts the average anesthesiologist at $543,000. Doximity's, published four months later, says $557,131. For family medicine, Medscape reports $288,000 and AMGA reports a median of $342,411, a 19% gap for one specialty in one year. None of the three is wrong. They count different people and report different statistics, and a physician who reads one number without knowing which is reading a figure that may not describe anyone in their own position.

Who is counted

The surveys split into two families. Medscape and Doximity ask physicians. MGMA, AMGA and SullivanCotter ask employers.

Medscape's 2026 report rests on 5,916 full-time U.S. physicians who answered an email invitation to its member base between September 5 and December 17, 2025. Doximity's 2026 report rests on about 23,000 surveys completed on its network during 2025; the same report drew 37,000 responses the year before, so the sample fell by roughly 38% in one year. Neither company verifies what a respondent reports.

MGMA's 2026 report draws on data that employers submitted for more than 245,900 physicians and advanced practice providers. AMGA's covers 451 medical groups and health systems and about 188,000 providers. SullivanCotter's 2026 physician survey covers 575 organizations and about 235,000 physicians. These figures come from payroll systems, so they are accurate for the people in them. The people in them are whoever works for an organization large enough to buy a benchmarking subscription and staff a survey submission every January. An HFMA analysis of a 2015 MGMA sample found one in seven providers practiced in Minnesota or Wisconsin, and New York did not appear in the top quintile of states.

No survey in either family weights its sample to the national physician population. A provider is a provider, so a health system that submits 8,000 physicians shapes a specialty's median far more than a four-physician practice that submits four.

What "compensation" includes

The word means something different in each report, and the definitions are where most of the gap between two numbers for the same specialty comes from.

  • Medscape 2026. What the pay figure includes: Base salary, incentive bonus, profit-sharing contributions. What it leaves out: Benefits, employer retirement contributions.
  • Doximity 2026. What the pay figure includes: Not defined in the 2023, 2024, 2025 or 2026 methodology sections. What it leaves out: Unknown.
  • MGMA 2025 definitions. What the pay figure includes: W-2, 1099 or K-1 direct compensation; bonuses and incentives; on-call pay; research stipends and honoraria; voluntary salary reductions such as 401(k) deferrals. What it leaves out: Expense reimbursements; employer-paid fringe benefits including retirement contributions and health insurance.
  • AMGA 2026. What the pay figure includes: Base plus variable clinical pay; on-call and supervision stipends; a second tier adds administrative and academic pay. What it leaves out: Employer retirement and pension contributions; prior-year signing and retention bonuses.
  • SullivanCotter. What the pay figure includes: Base and total cash compensation; benefits reported as a separate benchmark. What it leaves out: Pay for call coverage and extra work, which MGMA and AMGA include.

Two of these matter more than the rest. A hospital-employed surgeon with $60,000 of call pay appears $60,000 richer in MGMA than in SullivanCotter. And Doximity's figure cannot be compared with any of the others on this axis, because Doximity has never said what it is.

I have seen this move a comparison dramatically. Call can be written into a contract as part of the job, or guaranteed to the system, while the benchmark it is compared against counts call pay on its own line. Left unnegotiated, that difference can be worth more than $100,000 a year to the hospital.

Mean, median, or model

Medscape reports means, and its 2026 figures are partly modeled: the company says it estimated compensation from specialty, age, gender, location and the self-reported answers. It does not publish medians, percentiles, or the number of respondents in any specialty, and 5,916 physicians spread across 29 specialties leaves several specialties with thin cells.

Doximity reports an "average" that it never labels as mean or median, and the figure is regression-adjusted for specialty, metro area, gender, years in practice and hours worked. A regression-adjusted average is a modeled quantity. It is useful for comparing metros, which is what Doximity built it for, and less useful as an estimate of what a given physician takes home.

The employer surveys report percentiles, MGMA's from the 10th through the 90th, which is the form a compensation committee uses. One property of that form trips up readers. MGMA ranks each ratio separately, so the physician at the 90th percentile of total pay is not the physician at the 90th percentile of pay per work RVU. The relationship runs the other way: high producers tend to earn less per unit, because pay plans flatten above a threshold.

What a full-time physician is

MGMA's definitions state that a 1.0 FTE works "whatever number of hours the practice considers to be the minimum for a normal workweek, which could be 37.5, 40, 50 hours, or some other standard." AMGA's instructions say the same in different words, with 36 hours of patient care as an example. Medscape and Doximity ask respondents to self-classify; Doximity requires 40 hours a week. So the median for a specialty pools a 36-hour hospitalist with a 55-hour one, and the survey does not know which is which.

Even for employed physicians, I know physicians with 0.8 or 0.9 FTE contracts. This may give them the benefits of a full-time employee while also allowing for contracting work outside of their primary employment contract. Obviously, that can be tricky, but the key for this conversation is that income is not reflected by their primary contract.

When the number is from

Every 2026 report describes 2025 pay. Medscape fielded its survey from September to December 2025 and published in April 2026, and for nine specialties, plastic surgery among them, it also used data from its 2024 survey. Doximity collected through calendar 2025 and published in August 2026. MGMA's submission window runs January to mid-February, with publication in June. AMGA's deadline was March 31, 2026, with results released June 24. A physician reading any of these in the fall of 2026 is reading a number that is 9 to 21 months old, from before the 2026 fee schedule changes took effect.

The spread inside a specialty

The specialty average hides more variation than it reports. In Doximity's 2026 data, the gap between the highest- and lowest-paying metro areas, Oklahoma City at $495,617 and Durham-Chapel Hill at $381,424, is 29.9%. The gap by practice setting is wider: a single-specialty group averages $480,961 and government employment $311,502, a 54.4% spread, though Doximity does not say whether it held specialty mix constant across settings. Where a physician works moves the number more than where they live.

Individual-level data show the same thing from the other direction. Marit Health's May 2025 analysis of anonymized, moderated salary reports put the distance between a specialty's 10th and 90th percentiles at 21% to 72% of its median, with the widest spreads in fields where ownership is possible (ophthalmology, orthopedics) and some of the narrowest in hospital-based fields where pay is standardized and often salaried (anesthesiology, emergency medicine, hospital medicine). SalaryDr's 3,457 self-reported, unverified salary reports from 2024 to 2026 put the 90th percentile at 1.2 to 2.0 times the 25th within individual specialties. In each of the 18 specialties it lists, that spread is wider than the gap between the specialty's median and the next one's. Those sources are smaller and less audited than the employer surveys, so treat the ratios as the shape of the distribution rather than as the numbers.

What the number cannot tell you

None of the five reports says what a physician's pay will do next year. They are backward-looking by design.

None of them separates the terms that produce the number: what a payer pays per unit, how many units a physician produces, and what share of the collected dollar the employer passes through. Two physicians with identical survey pay can have opposite exposure to a reimbursement cut, depending on which of those terms the cut lands on.

How to read one

Find the methodology page before the table. Check whether the figure is a mean, a median or a model. Check whether call pay and bonuses are in it. Check the survey year, and subtract a year from the publication date. If the report offers a practice-setting cut, read that row before the specialty row. And if two reports disagree by 15% for the same specialty, assume both are right about their own samples before assuming either is right about you.

Where the book picks up

The Income Variable treats the compensation benchmark as one of three layers of information asymmetry between an employed physician and the institution that measures their value, and shows how the surveys the employers produce become the benchmarks the employers set pay against. The chapter titled "Death by a Thousand Cuts" puts that layer beside the payer and productivity layers it sits between. Join the launch list for the book and the free companion tools.

Sources

  1. Medscape, Physician Compensation Report 2026 (n=5,916; fielded Sept 5–Dec 17, 2025)
  2. WebMD press release, Apr 2025: Medscape 2025 report methodology (n=7,322, margin of error)
  3. Becker's Hospital Review, Apr 2026: Medscape 2026 methodology (5,916 physicians, Sept 5–Dec 17, 2025; 2024 survey data for nine specialties)
  4. Doximity, 2026 Physician Compensation Report and methodology
  5. Doximity, 2025 Physician Compensation Report (37,000 surveys completed in 2024)
  6. MGMA, 2025 Compensation and Payroll Definitions
  7. MGMA, DataDive FAQ (percentiles, ratio ranking, survey cycle)
  8. MGMA, Jun 10, 2026: 2026 Provider Compensation report release (245,900+ providers)
  9. AMGA, 2026 Medical Group Compensation and Productivity Survey instructions (definitions, FTE, confidentiality floor)
  10. AMGA, 2026 survey public charts (medians)
  11. AMGA, Jun 2026: 2026 Medical Group Compensation and Productivity Survey results release (451 groups, 188,000 providers)
  12. SullivanCotter, Physician Compensation and Productivity Survey
  13. SullivanCotter, Sep 2026: 2026 Physician Compensation and Productivity Survey results (575 organizations, about 235,000 physicians)
  14. KSM, Apr 2021: what is and is not in provider compensation survey data (call pay treatment; percentile misuse)
  15. HFMA, Jul 2018: geographic skew in the 2015 MGMA physician compensation survey sample
  16. Marit Health, May 2025: within-specialty pay spread (10th to 90th percentile as a share of the median)
  17. SalaryDr, Oct 2026: physician salary percentiles by specialty (3,457 self-reported, unverified reports, 2024–2026)
  18. MedPAC, March 2026 Report to Congress, Ch. 4 (physician compensation trends)

This article is general information and analysis, not individualized medical, financial, investment, tax, or legal advice. See the Disclaimers page for the full statement.

Most recent articles

Browse all articles
Robotic physician reading a personal finance book

The earning side is changing. Stay ahead of it.



Join the launch list for book updates, new articles, and free companion tools.

By submitting, you agree to our Terms of Use and acknowledge our Privacy Policy.