CONTEXT Data are sparse on the effect of varying the durations of internal medicine attending physician ward rotations. OBJECTIVE To compare the effects of 2- vs 4-week inpatient attending physician rotations on unplanned patient revisits, attending evaluations by trainees, and attending propensity for burnout. DESIGN, SETTING, AND PARTICIPANTS Cluster randomized crossover noninferiority trial, with attending physicians as the unit of crossover randomization and 4-week rotations as the active control, conducted in a US university-affiliated teaching hospital in academic year 2009. Participants were 62 attending physicians who staffed at least 6 weeks of inpatient service, the 8892 unique patients whom they discharged, and the 147 house staff and 229 medical students who evaluated their performance. INTERVENTION Assignment to random sequences of 2- and 4-week rotations. MAIN OUTCOME MEASURES Primary outcome was 30-day unplanned revisits (visits to the hospital's emergency department or urgent ambulatory clinic, unplanned readmissions, and direct transfers from neighboring hospitals) for patients discharged from 2- vs 4-week within-attending-physician rotations. Noninferiority margin was a 2% increase (odds ratio OR of 1.13) in 30-day unplanned patient revisits. Secondary outcomes were length of stay; trainee evaluations of attending physicians; and attending physician reports of burnout, stress, and workplace control. RESULTS Among the 8892 patients, there were 2437 unplanned revisits. The percentage of 30-day unplanned revisits for patients of attending physicians on 2-week rotations was 21.2% compared with 21.5% for 4-week rotations (mean difference, –0.3%; 95% CI, –1.8% to +1.2%). The adjusted OR of a patient having a 30-day unplanned revisit after 2- vs 4-week rotations was 0.97 (1-sided 97.5% upper confidence limit, 1.07; noninferiority P = .007). Average length of stay was not significantly different (geometric means for 2- vs 4-week rotations were 67.2 vs 67.5 hours; difference, –0.9%; 95% CI, –4.7% to +2.9%). Attending physicians were more likely to score lower in their ability to evaluate trainees after 2- vs 4-week rotations by both house staff (41% vs 28% rated less than perfect; adjusted OR, 2.10; 95% CI, 1.50-3.02) and medical students (82% vs 69% rated less than perfect; adjusted OR, 1.41; 95% CI, 1.06-2.10). They were less likely to report higher scores of both burnout severity (16% vs 35%; adjusted OR, 0.39; 95% CI, 0.26-0.58) and emotional exhaustion (19% vs 37%; adjusted OR, 0.45; 95% CI, 0.31 to 0.64) after 2- vs 4-week rotations. CONCLUSIONS The use of 2-week inpatient attending physician rotations compared with 4-week rotations did not result in an increase in unplanned patient revisits. It was associated with better self-rated measures of attending physician burnout and emotional exhaustion but worse evaluations by trainees. TRIAL REGISTRATION clinicaltrials.gov Identifier: NCT00930111