Coronary surgery is being used increasingly in older and more seriously ill patients. To evaluate the importance of age in predicting complications and mortality after coronary surgery, the results in 13,625 patients undergoing coronary surgery were evaluated. There were 321 patients under the age of 40, 1,758 were 40-49, 4,167 were 50-59, 5,049 were 60-69, 2,184 were 70-79, and 146 were over 80. As patients aged the proportion of women increased, the prevalence of severe angina and congestive heart failure increased, and the severity of the distributions of the coronary stenoses increased. The incidence of the complications of neurological events, wound infections, and death rose in every decade after the age of 40 years. The incidence of death was 0.6% for patients aged 40-49, 1.5% for those aged 50-59, 1.9% for those aged 60-69, 5.0% for those aged 70-79, and 8.3% for those aged 80-89. Advanced age was the most significant independent correlate of neurological events, wound infections, and death. The other multivariate correlates of neurological events were diabetes, hypertension, and, surprisingly, angina class. Again surprisingly, the only other multivariate correlate of wound infections was angina class. The other multivariate correlates of death were emergent surgery, female gender, ejection fraction, severity of coronary disease, diabetes, and a history of congestive heart failure. Of interest, periprocedural Q wave myocardial infarction was independent of age. The most predictive and significant correlate of periprocedural myocardial infarction was emergent surgery; hypertension, absence of a previous myocardial infarction, and a normal ejection fraction were weaker correlates. The ability to predict any complication was relatively weak compared with the ability to predict death. The relative infrequency of coronary surgery over the age of 80 years suggests that there is already strong selection against surgery in octogenarians. In addition, the risk of death, and to some extent other complications, may be predicted in advance based on extensive patient experience.