How Physician Compensation Surveys Work: MGMA, AMGA, and SullivanCotter Explained

Why survey data matters — and when you first hear about it

For most physicians, the first time someone cites a compensation survey in their career is during contract negotiations — and it is almost always the employer's recruiter or attorney doing the citing, not the physician. Understanding how these surveys are built, and what their numbers actually mean, puts you in a much stronger position before that conversation begins.

Physician compensation surveys are commercial reports that collect pay and productivity data from hospitals, medical groups, and health systems across the country, then publish the results broken down by specialty and geographic region. They serve two related purposes: they give physicians a market reference for evaluating whether an offer is competitive, and they give employers a defensible basis for setting compensation in a way that complies with federal fair market value requirements.

The three surveys that dominate the market

Three reports account for the overwhelming majority of citations in physician employment negotiations and formal fair market value analyses.

MGMA — Provider Compensation and Productivity Report

Published by the Medical Group Management Association, MGMA collects data from medical groups and health systems nationwide. The 2025 edition reflects 2024 calendar-year data from more than 220,000 physicians and advanced practice providers. It reports total cash compensation, work RVUs (wRVUs), and compensation per wRVU by specialty, with regional breakdowns. MGMA draws from a broad mix of practice types and sizes, making it one of the most widely cited single sources across specialties.

AMGA — Medical Group Compensation and Productivity Survey

The American Medical Group Association publishes an annual survey that, in its 2025 edition, drew from roughly 184,000 providers across nearly 500 medical groups spanning approximately 190 specialties. AMGA respondents tend to be larger integrated group practices and health systems rather than smaller independent groups — a distinction worth keeping in mind when comparing your situation to the data.

SullivanCotter — Physician Compensation and Productivity Survey

SullivanCotter's 2025 edition covers more than 232,000 providers across 500-plus organizations in 232 specialties, making it one of the broader surveys by specialty count and sample size. Hospital-employed practices represent a significant share of SullivanCotter respondents. For 2026, SullivanCotter expanded its specialty list to include cardiac surgery, neuroendovascular surgery, and primary care — rural medicine.

Each survey captures similar core dimensions — total cash compensation, wRVU productivity, and compensation per wRVU — but their respondent pools differ. The same specialty can show meaningfully different median figures across the three sources. Employers and valuation consultants typically consult multiple surveys rather than relying on one alone, which is also what CMS guidance recommends for fair market value analyses.

How to read a percentile table

Each survey reports compensation at several points across the distribution for each specialty. The most commonly cited points are the 25th percentile, the median (50th percentile), the 75th percentile, and the 90th percentile.

A single number without its percentile label is nearly meaningless. An employer quoting "the MGMA median for family medicine" is giving you the 50th-percentile total cash compensation figure — useful context, but not a complete picture of whether the specific mechanics of your offer (your wRVU threshold, your $/wRVU rate, your bonus structure) are individually fair.

Survey tables typically break out three metrics independently:

• Total cash compensation — base salary plus bonuses, before benefits.
• Work RVUs produced — productivity reported by the same specialty population.
• Compensation per wRVU — total cash divided by total wRVUs.

These three metrics are percentile-ranked independently of one another. That creates an important trap: a physician producing at the 75th percentile for wRVUs does not automatically earn 75th-percentile total cash compensation. Likewise, combining the high end of the comp-per-wRVU column with a high wRVU volume can produce a total compensation figure far above what the total cash column shows at the same percentile — a mathematical inconsistency that valuation analysts are trained to flag.

Fair market value and why it shapes what employers can offer

Physicians employed by hospitals or health systems operate under the constraints of the federal Physician Self-Referral Law, commonly known as Stark Law. The law is designed to prevent hospitals from using above-market compensation as an indirect payment for a physician's referrals to the hospital. Under the Stark Law employment exception (42 C.F.R. § 411.357(c)), total physician compensation must reflect fair market value and cannot be set in a manner that accounts for the volume or value of the physician's referrals.

In the 2020 Stark Law final rule, CMS explicitly rejected any automatic bright-line standard tied to a specific survey percentile. The rule makes clear that compensation above the 75th percentile is not automatically non-compliant, and compensation below the 75th percentile is not automatically safe. What matters is whether total compensation is genuinely supported by objective evidence — such as multiple independent surveys — given the physician's specialty, geography, experience, and scope of services.

The practical implications for your negotiations:

• If your employer tells you an offer is capped because of fair market value, that ceiling reflects their valuation team's reading of applicable survey data — not simply an opening position. Understanding which survey and percentile they are using gives you a factual basis for the conversation.
• Above-the-75th-percentile offers are possible but require documented justification from the employer, and often an independent FMV opinion letter from a qualified valuation firm. Rare subspecialty expertise, a critical community need, or a geographic premium can support a higher number — but the employer must be prepared to document it.
• Compensation per wRVU at an elevated percentile, combined with high wRVU volume, can produce a total compensation figure that exceeds what the surveys show at any reasonable percentile — a combination that valuation analysts examine closely. This is why a single metric in isolation does not tell the whole story.

Common mistakes when using survey data yourself

Before walking into a negotiation with printed percentile tables, recognize the most frequent misreadings:

• Using the wrong specialty bucket. Surveys subdivide specialties differently — "general internal medicine" in one survey may map to "hospitalist" or "internist — general" in another. The wrong bucket produces the wrong comparison, so verify you are referencing the correct specialty category before citing a figure.
• Mixing independent percentile columns. Presenting the 75th-percentile total cash figure alongside the 75th-percentile comp-per-wRVU figure from the same table overstates what the data supports, because each column is ranked independently. You cannot combine the high end of each column without creating an internally inconsistent picture.
• Ignoring regional variation. National figures are a starting point. Physician pay varies materially by region. Most surveys provide regional breakdowns; when available, use the region that matches your market rather than the national figure.
• Treating the median as a floor. The median reflects where the typical physician in your specialty lands, not a minimum you are entitled to. Whether an offer above or below the median is appropriate depends on your expected productivity, call burden, administrative responsibilities, and experience.
• Comparing medians across surveys without context. Because MGMA, AMGA, and SullivanCotter draw from different respondent pools, one survey's median may sit higher or lower than another's for the same specialty. An employer citing one survey's lower median and a physician citing another survey's higher median are both technically citing the median — just from different data sets.

Using survey data effectively in a negotiation

Survey data is most useful as a structured reference frame across the three compensation components that matter most in any physician employment offer:

1. Total cash compensation percentile. Where does the offer fall relative to the specialty distribution? Knowing the percentile lets you frame the discussion around market position rather than a general request for more.
2. Your wRVU threshold vs. the specialty productivity distribution. The threshold determines how much production you must generate before bonus pay begins. A threshold set at the 75th-percentile wRVU level, combined with median compensation per wRVU, is structurally disadvantageous — you must produce at a high level just to earn the median rate. Confirm where your threshold sits relative to the survey's wRVU productivity column for your specialty.
3. Your effective compensation per wRVU. Some employers state a $/wRVU rate explicitly; others embed it within a base-plus-bonus structure. Calculating your effective rate — projected total compensation divided by projected wRVUs — and comparing that figure to the comp-per-wRVU distribution lets you put both offer types on the same footing.

The most effective negotiating position is always a specific, documented argument: your specialty, your region, your expected productivity based on past output, and where those inputs land in the relevant distribution. That framing is considerably harder to dismiss than a general assertion that the offer should be higher.