To determine the preoperative variables affecting the mortality rate and the development of severe complications in patients who have had myocardial revascularization or a valve replacement and who then undergo a noncardiac operation, we retrospectively studied data from 120 such patients over the 5 years from 1982 through 1986. Thirty-six percent of patients had a noncardiac operation during the first month after the cardiac operation. The mortality rate was 11%, and the morbidity rate was 56%. The statistical comparison of the predictive accuracy of postoperative complications of three simple, widely used classifications (American Society of Anesthesiologists physical status, New York Heart Association classification, Massachusetts General Hospital cardiac risk index) demonstrated the superiority of the simplified three-class cardiac risk index (Massachusetts General Hospital-cardiac risk index; predictive accuracy of 84%). In a multivariate discriminant analysis of 21 variables in this population, five variables (myocardial infarction in previous 6 months, S3 gallop or jugular vein distention, arrhythmia on last preoperative electrocardiogram, emergency operation, delay between cardiac and noncardiac operation) were identified as being the most predictive of a postoperative complication. When these variables were used in the function (DF3) obtained by linear discriminant analysis, the prediction accuracy of a postoperative complication reached 83%. Performance of the new models in a prospective validation population remained satisfactory (75% for Massachusetts General Hospital-cardiac risk index three-class index and 72% for DF3). Extensive statistical analysis of our data tested by a validation study provided simple predictive models based on clinical variables easily available even in emergency situations.