OBJECTIVES--To study the risk factors for cardiac mortality after coronary artery bypass graft surgery. DESIGN AND SETTING--Follow up study of patients who had undergone coronary artery bypass graft surgery at the University Hospital of Oulu, Finland. PATIENTS AND INTERVENTIONS--339 consecutive patients who underwent cardiac catheterisation three months after bypass surgery. MAIN OUTCOME MEASURES--Incidence of cardiac deaths during the follow up period of five years and predictive value of clinical and angiographic variables for subsequent cardiac mortality. RESULTS--The incidence of cardiac deaths was 5.1%, and 81% of these were sudden deaths. The postoperative ejection fraction was significantly lower in the patients with subsequent cardiac death than in the survivors (p less than 0.001), and their left ventricular end systolic and end diastolic volumes were higher (p less than 0.001 and p less than 0.05 respectively). The incidence of cardiac deaths was 43% in the patients with a postoperative ejection fraction of less than 40%. The myocardial jeopardy index after surgery and the rate of graft patency were not significantly different in the survivors and patients who died. The only clinical factors that were different between the groups were postoperative use of diuretics (p less than 0.001) or digitalis (p = 0.02). After adjustment for other prognostic variables by the proportional hazards method, a low postoperative ejection fraction remained significant as a predictor of the relative risk of cardiac mortality five years after operation (p less than 0.01). CONCLUSIONS--Patients with angiographic evidence of impaired left ventricular function after bypass surgery are still at relatively high risk of dying suddenly, but myocardial ischaemia due to incomplete revascularisation is not strongly associated with an increased risk of cardiac mortality. Conventional clinical methods do not seem to be helpful for identifying patients with an increased risk of cardiac death after bypass surgery.