BACKGROUND: Noninvasive predictors of important outcomes after valve replacement for mitral regurgitation have not been examined in a rheumatic population (in whom the results of valve repair are suboptimal) in the era of chordal preservation. Timing of valve replacement thus remains a difficult question in rheumatic mitral regurgitation. METHODS AND RESULTS: Of 278 patients followed after valve replacement, 66 had pure or predominant mitral regurgitation, and in 61 of these the etiology was rheumatic. The mean age was 24 years. After a mean follow-up of 24 +/- 10 months, the ability of preoperative clinical and echocardiographic data to predict outcome was assessed prospectively, and the possible impact of chordal preservation (n = 35) on survival and post-operative left ventricular function was examined retrospectively. There were no perioperative deaths. There were six postoperative deaths, all the result of heart failure and all related to left ventricular dysfunction. The mean probability of survival was .90 at 16 months. In a stepwise Cox proportional hazards regression analysis, the only independent predictor of postoperative death was preoperative end-systolic diameter. According to a logistic model, the probabilities of death (n = 6) and death or severe heart failure (n = 7) increased abruptly at a preoperative end-systolic diameter of 51 mm (probabilities, .23 and .31, respectively), and the accuracy of this cut point for predicting outcomes was 97% and 98%, respectively. Multiple linear regression analysis identified a large preoperative end-systolic diameter and the need to use tricuspid annuloplasty as significant independent predictors of postoperative fractional shortening; the use of chordal preservation (n = 35) was not a predictor of postoperative fractional shortening. A good outcome was predicted at a preoperative end-systolic diameter of 40 mm: probability of death or heart failure was .0001, and predicted mean postoperative fractional shortening was 0.27 after mitral valve replacement without tricuspid annuloplasty. CONCLUSIONS: When preoperative end-systolic diameter is more than 50 mm, a poor postoperative outcome is predicted despite chordal preservation in relatively young patients with rheumatic mitral regurgitation, and alternative strategies should therefore be considered. When preoperative end-systolic diameter is 40 mm or less, an excellent outcome is predicted, and close observation without surgery would appear to be reasonable in the absence of symptoms.