$68,166 a year. Divide by the hours. The picture changes.

The average first-year stipend was $68,166 in the 2025 national survey covering 350 nonprofit teaching institutions and 114,361 residents and fellows^[1]. Stipend growth has trailed inflation.

Almost every hospital pays the same base across fields: 96.9% of institutions^[2]. The surgery resident at 83.1 hours a week and the psychiatry resident at 49.5 hours receive the same base check.

Inputs: $68,166 per year^[1]. Fifty-two weeks. Specialty hours from the 2025 survey of 479 residents and fellows^[3].

One Los Angeles chief resident estimated county residents earn less than $18 an hour^[28]. The math above supports that magnitude.

For scale: the federal minimum wage is $7.25 an hour^[19]. Residents earn more than that floor. They also hold medical degrees, carry life-and-death caseloads, and often owe hundreds of thousands in loans.

A medical degree costs dearly before residency pays anything. Four-year cost of attendance tops $286,000 at public schools and $390,000 at private schools (2024 figures; 2026 medians run $297,745 and $408,150)^[16]. The class of 2025 averaged $223,130 in total education debt including premed loans^[12]. Median medical-school debt was $200,000 with median premed debt of $28,000^[13].

Seven in ten graduates left school with debt, and 28% — about one in four — owed over $300,000^[14][15]. Tax relief is thin: the student-loan interest deduction caps at $2,500 a year with an income phaseout^[20]. About 90% of indebted graduates seeking forgiveness planned to use Public Service Loan Forgiveness^[21].

At Medicare's 1965 creation, Congress put training costs into teaching hospitals' “reasonable costs” — “including stipends of trainees” — so payments would “partially recompense salaries and benefits for house officers, administrative costs for GME, and cost of faculty”^[79].

Today's mechanism runs through two channels. In fiscal 2023 Medicare paid teaching hospitals an estimated $21.2 billion for training: $6.1 billion in direct payments plus $15.0 billion in indirect payments, supporting about 112,000 trainee slots^[80]. The average per-resident amount was about $133,000 for primary care and $131,000 otherwise, ranging from roughly $17,000 to $307,000 across hospitals^[81]. Direct payments equal a hospital-specific amount times trainee count times Medicare patient share, rooted in base-year costs “including teaching physician and resident salaries”^[82]. All federal training support runs about $29 billion^[83].

We state what this does not mean. The per-resident amount is not a stipend: it bundles supervision and overhead and covers only Medicare's share, so a $133,000 figure never means the resident is “paid” $133,000. Indirect payments target patient-care costs, not trainee pay. The money goes to hospitals, which set stipends; the pass-through is partial by design.

Resident stipends are taxable compensation. The IRS directs taxpayers to include in gross income “everything you receive in payment for personal services. In addition to wages, salaries…”^[84]. On payroll tax the IRS holds that “medical residents are not students… they are full-time employees,” ineligible for the student payroll-tax exception, so resident services since April 1, 2005 are subject to payroll tax^[85]. The student exception applies only where education predominates over employment^[86].

The asymmetry is plain: residents owe income plus payroll tax like employees while training under student-like constraints. We claim no deduction or exclusion advice beyond these sentences.

At 22 major nonprofit centers, mean chief-executive pay rose $1.6M to $3.1M (+93%) and finance-chief pay $740K to $1.4M (+83%) from 2005 to 2015, inflation-adjusted, drawn from tax filings^[92]. Gaps widened: chief executives went from 3 to 5 times orthopaedic-surgeon pay, 7 to 12 times pediatrician pay, and 23 to 44 times nurse pay^[93]. Nationally over the same span, healthcare mean wages rose 8%, management 14%, and physicians 10%^[94].

The pattern continued. From 2012 to 2019, nonprofit-hospital chief-executive mean pay rose 34% ($1.0M to $1.3M in 2019 dollars) while nurse mean wages rose 2.3% ($75,652 to $77,460)^[95] — and the link between chief pay and care quality weakened: executives are “rewarded more for leading large hospitals or systems, but not for providing higher quality care”^[96].

Stated limits: the papers compare site-specific executive pay to national clinician means, and the 22 centers are elite, not typical. Management growth partly reflects real regulatory burden. We claim no specific hospital underpays residents to fund its chief.

May 2025 national means from the Bureau of Labor Statistics: nurse practitioners $137,300 (median $132,300), physician assistants $141,280 (median $135,880), registered nurses $101,420 (median $97,550)^[97].

Ratios: $137,300 / $68,166 = 2.01×; $141,280 / $68,166 = 2.07×; $101,420 / $68,166 = 1.49× (all 2025)^[98]. Growth context: physician-assistant wages rose 40% from 2000 to 2013 against 35.3% cumulative inflation^[99]; from 2001 to 2017, nurses including NPs gained 9.92% cumulative median real growth, practitioners 5.68%, and physicians 37.6%^[100].

Stated limits: advanced practice clinicians are fully licensed earners with different training debt, hours, and liability — this is a level comparison, not an equal-work claim. The employment-weighted national means and the unweighted institutional resident mean make ratios approximate. No sourced clinician-hours figure exists, so we make no hourly comparison. The nurse-growth cluster includes NPs rather than isolating them.

Fairness includes the rules themselves. In 2003 the limits were 24 hours plus up to 6 additional (30 straight), call no more often than every third night, one day in seven free, and internal extra shifts inside the 80 (88 with exception)^[101]. The 2011 rules capped first-years at 16 hours, set others at 24 plus 4 (28, down from 30), counted all extra shifts inside the 80, and barred first-years from extra shifts^[102]. In 2017 first-years returned to 24 plus 4 — the “same schedule as other residents” — with the 80-hour week, one-in-seven, and every-third-night limits explicitly unchanged^[103].

Net across 23 years: the 80-hour week, every-third-night limit, and one-day-in-seven never moved. Only the intern and extended-shift cap traveled (30 to 16 to 28) plus a handoff narrowing from +6 to +4^[104]. Stability proves nothing in either direction: defenders read consensus, critics read capture. The 2017 reversal cites trial evidence — the one era change with trial backing.

A 2023 systematic review across 21 studies and 15,585 residents finds moonlighting common. About 25.2% of non-orthopedic residents moonlight, versus 10.3% of orthopedic residents^[50]. Debt drives much of it: about half of moonlighters in one study did so to pay debt, 31% cited debt in another, and larger debt predicted more moonlighting in a 537-resident study^[51].

Rules constrain it. All moonlighting counts toward the 80-hour limit, and residents in their first year may not moonlight at all^[52].

The other side: moonlighting also brings extra income, autonomy, and clinical repetitions some residents value. Program leaders worry about fatigue and drift from training goals. Prevalence varies widely by specialty. We state no pay rate and no net-harm verdict, since evidence covers motives more than outcomes.

Childcare can exceed a resident paycheck under a single-income model. The Family page works the math with clear scope.

Share the hourly math with one leader in your institution. Send the overtime illustration to a policymaker.