Training rules allow up to 80 hours a week, with shifts up to 28 hours.
The national training rules allow these schedules, and hospitals use them. Below: the rules, what residents report, duty-hour honesty, and what the evidence says about hours, errors, and safety — both sides shown.
What the rules allow
The national training rules cap clinical plus school work at 80 hours a week, averaged over four weeks[8]. Averaging can hide uneven weeks. Work done from home, including charting and pages, counts toward the 80 hours. Moonlighting also counts toward the limit[52].
Residents can work 24 hours of in-house call in a row, plus up to 4 more hours for handoffs and teaching[9]. After a 24-hour call, the rules require at least 14 hours free[10]. Residents must receive at least one day in seven free, averaged over four weeks[11].
| Rule | Limit |
|---|---|
| Clinical + school work | 80 hr/week, 4-week average |
| Longest scheduled call | 24 hr + up to 4 hr handoff = 28 straight |
| Free time after 24-hr call | at least 14 hr |
| Days off | 1 day in 7, 4-week average |
Source: ACGME Common Program Requirements (Residency), §§6.20–6.21. Moonlighting counts toward the limit; PGY-1 residents may not moonlight: UW GME Work Hours Policy (citing ACGME CPR 6.25).
What residents report
Self-reported hours vary by field. Almost every hospital pays the same base stipend across fields: 96.9% of institutions[2]. Schedules differ. Checks do not.
General surgery residents reported 83.1 hours a week in a 2025 survey of 479 residents and fellows[3]. Psychiatry residents reported 49.5 hours[4]. Most residents at Los Angeles County public hospitals report 80-hour weeks, per their union’s member survey[29].
Hospitals value long shifts for continuity and stamina. Critics respond that tired minds learn less and err more. Both views deserve a hearing, and the safety synthesis below gives each its evidence.
Sources: Resident Insider / Panacea survey · KFF Health News (LA County) · AMA / AAMC (96.9% uniform base).
Duty-hour honesty
Hour logs deserve honest discussion. In a national survey under the 2011 rules (n=6,202), 52.9% of residents reported some noncompliance and 42.9% said they had falsely reported hours[47]. Surgery reported the highest noncompliance at 67.6%, and program directors independently reported 57.3% noncompliance among 549 respondents at a 75.2% response rate[48]. An earlier yearlong study found 83.6% of 1,068 interns violated hours in at least one month[49]. Junior residents violated more than seniors.
The other side matters. The authors note some violations reflect patient-care ethics overriding clock rules. Residents who disliked the rules violated more, so noncompliance is not pure coercion. The survey had a modest response rate, so nonresponse bias is possible. These figures describe 2011–2012 rules and cannot stand in for today. We name no program and assert no current rate.
The lesson for leaders: hours systems need simple counting, safe reporting, and steady audit. Accurate logs protect residents and programs alike.
Source: Drolet et al., JGME 2013 (PMC3771164), citing Landrigan et al. 2006 for the intern study.
Do long hours harm patients? A balanced synthesis
Where the evidence holds: errors and lapses
The Harvard Work Hours group randomized 20 ICU interns to 24-hour versus 16-hour schedules. Interns on long schedules made 35.9% more serious medical errors (136.0 vs 100.1 per 1,000 patient-days, P<0.001), across 2,203 patient-days and 634 admissions[66]. The gap widened on the errors that matter most: 56.6% more nonintercepted serious errors, a 22.0% higher unit-wide error rate, 20.8% more serious medication errors, and 5.6 times the serious diagnostic errors[67]. This trial offers the strongest causal read on real errors with blinded review.
The companion paper explains the mechanism. The short-schedule interns worked 19.5 fewer hours a week, slept 5.8 more hours a week, and showed less than half the attentional-failure rate on on-call nights[68]. Sleep loss cuts clinical performance by more than 1.5 standard deviations in a 60-study meta-analysis[71].
Harm extends beyond the ward. After extended shifts of 24 hours or more, interns faced 2.3 times the crash odds and 5.9 times the near-miss odds while driving; each added extended shift per month raised crash risk 9.1%[69]. Needlestick odds ran 1.61 times higher after extended work, with lapse in concentration (64%) and fatigue (31%) the top contributing factors[70].
Distress and errors travel together
Depressed pediatric residents made 6.2 times the medication errors per resident-month of their non-depressed peers (1.55 vs 0.25, P<0.001)[72]. Internal medicine residents screening positive for depression had 2.56 times the odds of a subsequently reported major error[73]. Nationally, burnt-out physicians had 2.22 times the odds of a self-reported major error[74].
Honest limits: in the pediatric study, the burnout-to-error link alone was null (P=0.2), so we claim no burnout-only effect[72]. The national figure covers attending physicians with self-reported errors, not resident-measured errors[74]. The Mental Health page carries the full picture on trainee distress.
The mortality caveat: where reforms did not move the number
We report this side in full. The FIRST trial randomized 117 surgery programs covering 138,691 patients: death plus serious complications ran 9.1% under flexible hours versus 9.0% under standard rules, meeting noninferiority[75]. Flexible-regime residents left operations less often (7.0% vs 13.2%) and handed off less (32.0% vs 46.3%)[75]. The iCOMPARE trial across 63 internal medicine programs found the change in 30-day mortality noninferior within a one-point margin[76]. A single-center study found no significant differences in length of stay, 30-day readmission, ICU admission, inpatient mortality, or hospital-acquired conditions in the first year after the 2011 reform[78]. A survey of 2,323 interns found self-reported serious errors ticking from 20% to 23% after 2011, with handoffs and work compression blamed[65].
Read these trials for what they are: duty-hour regimes show no measurable mortality difference inside the tested 80-hour-week envelope. They compare flexible versus standard caps, never capped versus uncapped hours, and mortality is too blunt an endpoint to catch the error and lapse harms the ICU trial and cohort studies do measure.
Trainee suffering itself is measurable. Flexible-regime interns were 2.47 times more likely to report dissatisfaction with well-being, and missed the licensing-exam noninferiority bar[77].
Honest synthesis
Our thesis is pitched where the evidence holds: long shifts raise errors and lapses under direct observation, raise crashes and needlesticks at national scale, degrade cognition under sleep loss, and deepen trainee suffering. System reforms have not clearly moved mortality, plausibly because handoffs, work compression, and coverage absorb gains. We claim no patient death count from duty hours, no guaranteed safer care from shorter shifts, and no endorsement of unlimited hours from trials that kept the weekly cap.
Sources: Landrigan et al., NEJM 2004 (PMID 15509817) · Lockley et al., NEJM 2004 (PMID 15509816) · Barger et al., NEJM 2005 (PMID 15647575) · Ayas et al., JAMA 2006 (PMID 16954484) · Philibert, Sleep 2005 (PMID 16335329) · Fahrenkopf et al., BMJ 2008 (PMC2258399) · West et al., JAMA 2009 (PMID 19773564) · Tawfik et al., Mayo Clin Proc 2018 (PMID 30001832) · Bilimoria et al., NEJM 2016 (PMID 26836220) · Silber et al., NEJM 2019 (PMID 30855740) · Desai et al., NEJM 2018 (PMID 29557719) · Block et al., J Hosp Med 2014 (PMID 24677678).
Support a real hours limit
Share this synthesis with a program leader. Invite them to compare the two evidence columns side by side.