Background. The surgical results of the Cox-Maze III procedure (CM-III) for atrial fibrillation (AF) associated with rheumatic mitral valve (MV) disease are not as good as the results from surgery for AF alone. Methods. To assess the efficacy and safety of the CM-III in AF associated with rheumatic MV disease, we retrospectively analyzed 75 patients who underwent the CMIII combined with a rheumatic MV procedure between April 1994 and December 1997. Fourteen cases were reoperations because of prosthetic valve failure. Results. Mean aortic cross-clamp (ACC) times and cardiopulmonary bypass (CPB) times were 151 ± 43 and 251 ± 73 min, respectively. Concomitant procedures were mitral valve replacement (MVR) in 25 patients, MVR and aortic valve replacement (AVR) in 14 patients, MV repair in 10 patients, MVR and tricuspid annuloplasty (TAP) in 6 patients, MVR and AV repair in 3 patients, MVR and coronary artery bypass grafting (CABG) in 2 patients, MVR and AVR and CABG in I patient, redo-MVR in 8 patients, redo-MVR and TAP in 4 patients, and redoMVR and redo-AVR in 2 patients. There were two in-hospital mortalities (2 of 75, 2.7%). Seventy-three survivors were followed for a mean duration of 30 ± 13 months (12-56 months). Normal-sinus rhythm was restored in 90.4% (66 of 73). Three patients remained in AF and 2 patients were in junctional rhythm. Permanent pacemakers were implanted in 2 patients due to sick sinus syndrome. Right atrial (RA) contractility was demonstrable in 100% (66 of 66) and left atrial (LA) contractility in 62.1% (41 of 66) of the patients in the latest follow-up echocardiography. RA and LA contractilities were restored a mean 69 ± 93 and 126 ± 136 clays after the operation, respectively. LA contractility was restored significantly later at a lower rate than RA contractility in rheumatic MV disease. There were no differences in ACC time, CPB time, incidence of postoperative bleeding complications, and sinus conversion rates between nonredo and redo cases in spite of the significantly longer duration of preoperative AF in redo cases (p < 0.05). Conclusions. The CM-III for AF associated with rheumatic MV disease demonstrated a high sinus conversion rate with acceptable operative risk even in cases of reoperation.