Decision-analytic techniques were used to evaluate the choice between an aortocoronary bypass operation and medical management in a set of hypothetical patients with coronary artery disease. The decision framework incorporates variables believed to have an important bearing on the choice of treatment. Probability estimates were obtained from two cardiologists and one cardiac surgeon. Patient preferences for the trade-off between years of survival and the quality of life as reflected by the severity of angina pectoris were made explicit by assigning utility values to alternative health outcomes. The results are expressed in terms of quality-adjusted years of life expectancy. Decision analysis favored operation for 13 of the 14 hypothetical patients, including patients with one- and two-vessel disease. The one patient for whom medical treatment was preferred had mild angina pectoris, severe left ventricular dysfunction, and a poor prognosis regardless of therapeutic modality. The results are sensitive to changes in the probability of long-term survival, but not to changes in operative mortality rates. In five patients, the physicians' clinical judgments favored medical treatment, whereas their decision-analysis-derived estimates of survival favored operation. Possible explanations for these discrepancies are discussed. A simplified cost-effectiveness analysis for patients in whom surgery was the optimal treatment indicated costs ranging from $1,500 to $250,000 per year of life gained and from $1,500 to $32,000 per quality-adjusted year of life gained.