Resident supervision by faculty is a sine qua non of surgical education, yet objective standards for supervision are difficult to quantify. Over a 12-month period, using departmental data on morbidity, mortality outcome, and faculty status in the operating room, the association between complications, death, and attending physician presence were analyzed by Chi-square tests of association in 2 x 2 contingency tables, or by the Mantel-Haenszel Chi-square to control for a stratifying variable. A total of 4417 cases were reported. Attending physicians were either scrubbed or present in the OR 91.8% of the time, although there was considerable variation among services. The overall mortality rate was 6.2% and complications occurred in 7.0% overall. Greater attending physician presence was significantly associated with lower mortality and complication rates overall. When stratified by service, the association was less marked. However, presence of attending physicians varied significantly by service. To adjust for this variation, elective services were compared with all the "nonelective" services. When this categorization was used as the stratifying variable, the association between increased attending physician involvement and decreased complication and mortality rates was statistically significant (Mantel-Haenszel Chi-square, p < 0.0005 for both).