Debate exists ove the most appropriate form of treatment for patients with unstable angina pectoris. This study examined 106 patients randomized at the University of Alabama in Birmingham as part of the National Cooperative Study Group and focuses on the phenomenon of patients who fail medical therapy and thus require late surgery, and the costs of therapy. Discriminant function analysis revealed that the significant predictors (p<0.01) of patients who would later require surgery were: total number of vessels diseased, angina severity, presence of congestive heart failure, hypertension, and number of years that the patient had had angina. By means of this analysis, 85% of the late surgery patients were correctly predicted. Late aurgery patients averaged 2.4 diseased vessels vs 1.5 for persistent medical patients (p<0.01). Mean charges for the first 2 years in the study were $6,226 (SD $2,967) for persistent medical patients, $10,416 (SD $2,146) for surgery patients, and $20,059 (SD $10,748) for late surgery patients (p<0.001). These data indicate that surgery is clearly an expensive procedure; but it is more expensive for late surgery patients, who have total costs that twice as high as surgical costs and 3.5 times as high as persistent medical costs.