Between 1976 and 1986, 19 children aged 1 month to 5 years underwent replacement of the mitral (systemic atrioventricular) valve. Indications for valve replacement included isolated congenital mitral stenosis (n = 2), valve dysfunction associated with a more complex procedure (n = 15), and failed valvuloplasty (n = 2). Seven different valve types were used; nine were mechanical valves and ten were bioprosthetic valves. There were 6 hospital deaths (32%; 70% confidence limits, 20% to 47%). Among the 13 survivors there were 3 late deaths at a mean of 14 months after operation. The late deaths were unrelated to valve malfunction. Thromboembolic events occurred in 2 patients, both with mechanical valves. One minor bleeding complication occurred among 10 patients on a regimen of Coumadin (crystalline warfarin sodium). Five patients, all with bioprostheses, required a second valve replacement. Indications for reoperation included prosthetic valve regurgitation (n = 1) and calcific stenosis (n = 4). No early or late deaths occurred after second valve replacement. Survival was 51% +/- 12% (standard error) at 112 months after valve replacement. Analysis failed to identify age, weight, sex, previous operation, underlying cardiac lesion, or prosthesis size and type as significant risk factors for mortality. Mechanical valves had a lower reoperation rate compared with bioprostheses. These data suggest that although mitral valve replacement within the first 5 years of life is associated with a high operative and late mortality, satisfactory long-term palliation for many patients can be achieved. Mechanical valves are superior to bioprosthetic valves, and offer the best long-term results.