The effect of medical and surgical treatment on subsequent sudden cardiac death was assessed in 13,476 patients in the Coronary Artery Surgery Study registry who had significant coronary artery disease, operable vessels, and no significant valvular disease. (Patients were assigned to medical or surgical therapy on the basis of clinical judgment and not according to a randomization scheme; therefore, biases associated with unknown variables could not be evaluated.) Sudden cardiac death occurred in 452 patients (3.4%) during a mean follow-up of 4.6 years. Five year survival free of sudden death for medically treated patients was 94 +/- 0.3%, and that for surgically treated patients was 98 +/- 0.2% (p less than .0001). Twelve baseline clinical, electrocardiographic, and angiographic variables were significantly different between patients alive at the last follow-up and those suffering sudden death. Data on these variables were available for 11,508 patients. Sudden death occurred in 257 (4.9%) of 5258 medically treated and 101 (1.6%) of 6250 surgically treated patients. In a high-risk patient subset with three-vessel disease and history of congestive heart failure, 91% of surgically treated patients had not suffered sudden death compared with 69% of medically treated patients. After Cox survival analysis was used to correct for baseline variables, surgical treatment had an independent effect on sudden death (p less than .0001). This reduction was most pronounced in high-risk patients.