OBJECTIVE The clinical evaluation and comparison of St. Jude Medical (SJM) and CarboMedics (CM) prosthetic heart valves implanted between 1988 and 1997 is presented. METHODS In total, 648 SJM valves were implanted in 641 patients, and 601 CM valves in 591 patients. There were 684 mitral valve replacements, 256 aortic valve replacements, 252 mitral and aortic (double) valve replacements, 16 triple valve replacements, and 41 other tricuspid-related valve replacements. Total follow up was 98%. The overall incidence of valve-related events was compared before and after establishment of a 'valve clinic' in 1993. RESULTS The overall hospital mortality was 3.4%; late mortality was 8.2%. The five- and ten-year survival for all patients was 92.1% and 86.2%, respectively. There were 31 episodes of thromboembolism in 27 patients (including valve thrombosis in three), 21 episodes of bleeding events in 20 patients, and 18 re-replacements of implanted valves. No structural valve deterioration was observed. Freedom from thromboembolism was 97.8% at five years and 96.3% at ten years; freedom from bleeding episodes was 98.1% and 97.6%, respectively. In terms of hospital and late mortality, and incidence of thromboembolism, hemorrhagic episodes and structural valve failure, no statistically significant differences were found between the SJM and CarboMedics patient groups. Freedom from thromboembolism was 96.7% at five years before initiation of an intensive follow up program, and 99.0% thereafter (p = 0.031). In contrast, freedom from bleeding episodes fell from 99.3% to 96.1% during the same time period (p = 0.0004). CONCLUSIONS Both the SJM and CM prosthetic heart valves performed well in our study, and no discernible differences in clinical performance of the two valves were detected. The intensive follow up program resulted in a reduced incidence of thromboembolism, but an increased number of bleeding complications. An optimum anticoagulation regimen to manage these two conflicting problems has yet to be elucidated.