Seminal studies and widely publicized cases have linked fatigue among trainee physicians with medical errors. In response, the Accreditation Council for Graduate Medical Education (ACGME) has progressively limited duty hours for residents over the past decade. While first-year trainees now may work no more than 16 consecutive hours, upper-level residents may still be on duty for 24 consecutive hours; the ACGME strongly recommends protected sleep time during such extended shifts. Conducted at a single internal medicine residency's two teaching hospitals, this randomized controlled trial found that residents assigned to receive protected sleep time did sleep more and were less fatigued—by subjective and objective measures—compared with residents who had no protected sleep opportunity. However, increased sleep did not translate into improved patient-level clinical outcomes, and extra staffing was required (at one of the two hospitals) to implement the intervention. As prior studies of earlier regulations also did not find improvement in clinical outcomes after duty hour reduction, the relationship between physician work hours and patient safety remains complex and poorly defined.