The effect of saphenous vein bypass grafting on the incidence of sudden cardiac death and total cardiovascular death was studied by comparing the survival of subsets of surgically and medically treated patients with coronary disease who were similar in two major prognostic variables: extent of coronary disease and ejection fraction. Significant differences in some baseline variables existed between medical and surgical subgroups with similar ejection fraction and extent of coronary disease. Medically treated patients tended to be in a lower functional class, to have more extensive ventricular contraction abnormalities, to have a larger end-diastolic volume, and to have fewer distal vessels feasible for grafting. Surgically treated patients with two vessel disease and normal or moderately reduced election fraction had improved survival when compared with medically treated patients with two vessel disease and similar ejection fraction. Improved survival of borderline statistical significance was also seen in surgically treated patients with three vessel disease and moderately reduced ejection fraction. When categorized according to end-diastolic volume, surgically treated patients with two vessel disease and normal or moderately increased end-diastolic volume also had improved survival over similarly defined medically treated patients. The sudden death rates for subgroups of mecically treated patients were 1.8 to 10.9 times higher than the rates in subgroups of surgically treated patients with a comparable extent of coronary disease and ejection fraction.