Mitral repair is feasible for patients with degenerative or ischemic heart disease, however, the appropriateness of repair for rheumatic heart disease remains controversial. We compared our outcomes for primary isolated mitral repair versus replacement in an elderly population. From November 1997 to July 2005, mitral repair (group I) was performed in 33 patients while 59 underwent replacement (group II). Survival and risk factors were evaluated by Kaplan-Meier and Cox regression analysis. Mean age at operation for groups I and II was 49.7 +/- 13.2 versus 58.1 +/- 11.2 (P = 0.002). No statistically significant differences with regards to demographic parameters were observed except for there being fewer percutaneous transvenous mitral commissurotomy procedures and a lower severity of pulmonary hypertension in group I. Patients with a greater Wilkins score and more valvular calcification underwent replacement more often (P < 0.001). In-hospital mortality, ICU/hospital stay, and postoperative congestive heart failure functional class did not differ significantly. Major adverse cardiac events occurred in 13 and 19 patients, respectively (P = 0.50). There were 4 versus 6 late deaths (P = 0.74). Only two from group I underwent subsequent mitral reoperation. Kaplan-Meier overall survival and event-free survival at 5 years for groups I and II were 0.81 +/- 0.08 versus 0.81 +/- 0.06 (P = 0.90) and 0.52 +/- 0.10 versus 0.51 +/- 0.10 (P = 0.21), respectively. Old age, renal insufficiency, LVEF < 40%, and a history of stroke were poor predictors of patient survival. Compared with replacement, mitral repair for rheumatic heart disease was associated with a lower surgical mortality, higher repeat-surgery rate, and good survival. Rheumatic mitral valves should be repaired in select patients with appropriate valvular pathology.