265 patients (168 women, 97 men) after cardiac surgery (mitral valve replacement--74 pts, mitral commissurotomy--158 pts, aortic valve replacement--6 pts, replacement of both valves--16 pts, closure of ASD--11 pts) were randomized after successful electro-conversion of atrial fibrillation to quinidine (63 pts), verapamil (56 pts), amiodarone (50 pts), digoxin (56 pts) or control group (40 pts). The groups were comparable regarding age, sex and mitral valve disease distribution, heart volume, echocardiographic left atrium size and time from cardiac surgery to electro-conversion. After one year sinus rhythm was still present in 43% receiving quinidine, 43% receiving verapamil, 40% receiving amiodarone, 22% receiving digoxin, 20% in the untreated group, and after two years in 14%, 11%, 20%, 0% and 0%--respectively. The treatment was discontinued because of side effects in 13% of pts in the quinidine group, 8% of pts in the amiodarone group and 4% of pts in the verapamil group. It is concluded that quinidine, amiodarone and verapamil compared with control group are significantly (p less than 0.05--after one year) more effective in preventing late relapses of atrial fibrillation. Digoxin is ineffective in preventing recurrence of the arrhythmia. There are no significant differences between quinidine, verapamil and amiodarone regarding its prophylactic efficacy.