Ninety patients undergoing coronary bypass surgery were studied prospectively by bedside and subsequent ambulatory electrocardiographic monitoring to investigate the incidence, possible causes, and prevention of atrial fibrillation. Patients with good left ventricular function were divided randomly into a control group or groups treated with digoxin or propranolol. In the control group the incidence of atrial fibrillation was 27% and of significant ventricular extrasystoles 3%. Propranolol reduced the incidence of atrial fibrillation (14·8%), whereas digoxin had no effect and increased the incidence of ventricular extrasystoles. Age, sex, severity of symptoms, cardiomegaly, heart failure, previous myocardial infarction, and number of grafts did not affect the result. The operative myocardial ischaemic time was related to the occurrence of atrial fibrillation. There was also a significant relation between atrial fibrillation and bundle branch block. Atrial fibrillation is common after coronary artery grafting; it may be due to diffuse myocardial ischaemia or hypothermic injury. The incidence may be reduced by beta blockade.