To test the contention that the results of cardiac surgery are better in recent years, we analyzed patients undergoing isolated or combined mitral valve replacement in an earlier era (1975 to July, 1979; n = 478) and a later era (July, 1979, to July, 1983; n = 341). Patients in the later era were older, had a higher left ventricular end-diastolic pressure, and a higher prevalence of ischemic mitral valve disease and a lower one of rheumatic mitral disease. Patients operated on in the later era had, on the average, longer and more extensive operations. Patients in the later era had a slightly but inconclusively lower 2 week and 4 1/2 year survival rate than those in the earlier era; their hazard functions were similar. A higher proportion of the deaths in the later era than in the earlier era were due to chronic heart failure. However, neither the era nor the specific year in which a patient was operated on was a risk factor for death in a multivariate analysis. By contrast, in many subsets of patients with congenital and ischemic heart disease, the risks have been shown to be lower in the current era. Continuing delay in advising mitral valve replacement and imperfections in the methods for myocardial protection are probably responsible for this lack of improvement across time.