Recent studies have suggested that patients with three-vessel coronary disease and abnormal left ventricular function have better survival rates with bypass surgery than with medical therapy alone. Case-control studies may give accurate survival estimates, but to be valid, selection biases must be taken into account. A matched case-control method was used to compare survival patterns in patients treated medically or surgically during the 1980s. Fifty medical patients with potentially operable coronary disease and 46 surgical patients were matched for significant three-vessel disease and abnormal ventricular function. These two groups had no significant differences with regard to 24 variables, including age (64 +/- 8 versus 63 +/- 10 years), chest pain class, congestive heart failure signs, ejection fraction (36 +/- 8 versus 37 +/- 9 percent), segmental wall score, or a coronary score evaluating lesion site and severity. There were slight differences between the two groups with regard to congestive heart failure symptoms (p = 0.04). Patients undergoing bypass surgery had improved four-year survival rates compared with the medical group (89 versus 55 percent; p = 0.01). Thus, this study used an effective case-control method to suggest that, in the 1980s, coronary surgery improves prognosis substantially in surgically approachable patients with severe coronary disease and ventricular dysfunction.