During 1972 and 1973, a total of 235 patients had open heart surgery for mitral valve disease unassociated with significant aortic or rheumatic tricuspid valve disease. Thirty-one underwent closed heart mitral commissurotomy, without mortality. Of the 204 patients undergoing open operation, 125 had sequential measurement of cardiac output and mixed venous oxygen pressure. The hospital mortality rate was 6.4 percent in the larger group of 204 patients and in the 125 with cardiac output measurements. The rate was greater in those with class IV disability (New York Heart Association criteria) preoperatively than in those with class III or II disability. The mean ± standard deviation of the average cardiac index early postoperatively was 2.05 ± 0.579 liters ·min-1·m-2. Cardiac index was lower in the patients who died early postoperatively than in those who did not. The probability of hospital death was a significant function of cardiac index. The predicted probability of death was 10 percent with an average cardiac index of 1.42 liters·min-1·m-2 and increased sharply with lower indexes. Cardiac index was lower early postoperatively than preoperatively, and was lower in patients in class IV than in those in class III. There was no significant difference in cardiac index between patients with mitral valve replacement and those with repair. A history of closed commissurotomy, age, duration of cardiopulmonary bypass, duration of cardiac ischemia and method of myocardial preservation did not significantly influence cardiac index or hospital mortality rate. There was no significant relation between mixed venous oxygen pressure and hospital death. Further improvement in results of mitral valve surgery requires adequate preservation of left ventricular performance before, during and after operation.