Objective – Report clinical experience in surgical treatment of atrial fibrillation (AF) by Cox-maze procedure. Methods – 61 patients underwent surgical treatment for AF. Two had primary AF and 59 AF secondary to heart disease (2 atrial septal defects, 57 mitral). Ages ranged from 20 to 74 years (mean = 49). There were 44 females (72%). The surgical technique employed was Cox 3 without cryoablation. The patients were follow-up in specific at patient clinics and underwent periodical ECG, exercise tests, echocardiogram and Holter monitoring. Results – In-hospital mortality was 4.9% and late mortality 1.6%. A temporary pacemaker was used in 28 (46%) and a definitive in 7 patients (11.4%). On hospital discharge, AF remained in 17%; 63.9% had sinus rhythm, 6.9% atrial rhythm, 1.7% junctional rhythm, and 10.3% had pacemaker rhythm. In the last evaluation, AF was present in 19.5%; (70.5% sinus rhythm, 4% atrial rhythm, 2% atrial tachycardia, and 4% pacemaker rhythm). There was no report of thromboembolic episodes. Chronotropic response was considered adequate in 19%, intermediate in 29%, and inadequate in 42%. In Holter monitoring, the mean heart rate was 82±8 bpm, with a minimum of 57±7 bpm and maximum of 126±23 bpm, with supraventricular extrasystoles in 2.3±5.5% of the total heartbeats and ventricular extrasystoles in 0.8±0.5%. In the echocardiogram, the A wave was present in the left atrium in 87.5%. Conclusion – Maze procedure is effective and has acceptable surgical risk. Atrial or sinus rhythms remain stable with a small but remarkable frequency of atrial and ventricular arrhythmias. Left atrial contraction is present, although attenuated, as well as the chronotropic response to exercise.