This study was authorized by the Department of Veterans Affairs to improve the quality assurance of cardiac surgery by assessing preoperative risk factors and relating them to operative mortality. Data were received on 10,480 patients over a 2-year period. Preoperative risk variables were subjected to univariate and multivariate logistic regression analyses. Significant variables for coronary artery bypass grafting after logistic regression analysis in order of importance are previous cardiac operation, priority of operation, New York Heart Association functional class, peripheral vascular disease, age, pulmonary rales, current diuretic use, and chronic obstructive pulmonary disease. For patients undergoing valve or other cardiac operations with or without coronary artery bypass grafting, those variables found to be significant after multivariate logistic regression analysis are priority of operation; age; peripheral vascular disease; great vessel repair; all other except aortic valve replacement, mitral valve replacement, and great vessel repair; mitral valve replacement; and cardiomegaly. By identifying these current risk factors and the coefficients from the multivariate stepwise logistic regression analysis, expected mortality can be calculated. We propose that the ratio of observed to expected mortality is a better measure of quality of care than unadjusted mortality.