As part of a prospective program to use risk-adjusted outcome (operative mortality and morbidity) as a measure of quality of care, we have analyzed perioperative complication data in 10,634 patients representing 73% of all patients undergoing cardiac surgery requiring cardiopulmonary bypass at Veterans Administration medical centers between April 1, 1987, and March 31, 1989. One or more complications occurred in 15% of patients undergoing coronary artery bypass grafting, and in 24% of patients undergoing valve and other cardiac surgery. Patients experiencing one or more complications had an eightfold to 10-fold increase in operative mortality compared with patients with no perioperative complications. The most frequent complication was requirement for mechanical ventilation for at least 48 hours occurring in 8% of patients undergoing coronary artery bypass and in 15% of patients undergoing valve and other cardiac surgery; 24-25% of these patients died within 30 days of surgery or as a direct result of a surgical complication. Previous heart surgery was a strong predictor of development of one or more complications in both groups of patients, being associated with an adjusted relative risk of 1.6-2.0. Other important predictors in both surgical groups were surgical priority, older age, peripheral vascular disease, and higher serum creatinine. Although a number of preoperative risk factors could be identified for the development of renal failure, low cardiac output, and requirement for prolonged mechanical support, few risk factors could be identified for the development of mediastinitis and reoperation for bleeding. This observation suggests that mediastinitis and reoperation for bleeding are more likely the result of technical factors rather than patient-related risk factors.