Our aims were to examine whether the administration of amiodarone or magnesium sulphate after coronary artery bypass graft surgery (CABG) could reduce the occurrence of atrial fibrillation, and to identify the risk factors associated with atrial fibrillation after CABG. Patients scheduled for elective CABG (n = 155) were allocated randomly, in a controlled double-blind study, to receive immediately after surgery a 72-h infusion of amiodarone (900 mg per 24 h), magnesium (4 g per 24 h) or placebo (0.9% NaCl; 50 ml per 24 h) intravenously. A 72-h Holter ECG was recorded concomitantly. The primary end-point was the prevention of atrial fibrillation; its onset was considered as prophylactic failure. An interim safety analysis was performed in 147 patients. The cumulative occurrence of atrial fibrillation was 27% in the placebo group, 14% in the amiodarone group (P = 0.14) and 23% in the magnesium group (P = 0.82). Although amiodarone delayed the onset of the first tachyarrhythmic episode (P = 0.02), it was associated with the need for longer periods of vasoactive drug infusion and invasive monitoring and a longer stay in the intensive care unit. Variables associated with the onset of atrial fibrillation were older age (odds ratio 1.9) and a plasma magnesium concentration at 24 h of less than 0.95 mmol litre-1 (odds ratio 6.7). Postoperative administration of amiodarone reduced the occurrence of atrial fibrillation after elective CABG surgery, but was associated with a longer duration of cardiovascular instability and longer need for intensive care; magnesium prophylaxis had no effect. Advanced age and a low plasma magnesium concentration are risk factors for postoperative atrial fibrillation.