The timing of operation and the selection of prosthesis depend upon the evaluation of long-term results at 5 years and beyond. From 1965 to 1975, 290 patients had isolated mitral valve replacement with currently used prostheses, resulting in a 6.6% operative mortality and 64% relative survival at 10 years. In considering mortality, the results obtained with both series of valves were combined and a search made for significant differences in preoperative variables between the operative deaths and operative survivors. The significant variables for operative mortality were valve model, preoperative functional class, pump time, and the presence or absence of right ventricular hypertrophy on electrocardiography. There were significant differences in long-term survival only with regard to preoperative functional class, pulmonary vascular resistance, and pump time. Exponential curves were fitted to the event-free rates, and long-term results were then extrapolated. The 10-year projection of the percentage of patients alive and free of embolus (disregarding transient ischemic attacks) for the Model 6310-6320 (50%) was similar to that actually observed for the Model 6120 (46%). Although further experience may show one prosthesis more effective than the other, statistical extrapolations suggest continued comparable performance.