From 1975 to 1979, 540 patients undergoing valve replacement were entered into a randomized trial and received either a Björk-Shiley (273 patients) or a porcine heterograft prosthesis (initially a Hancock valve [107 patients] and later a Carpentier-Edwards prosthesis [160 patients]). Two hundred and sixty-two patients required mitral valve replacement, 210 required aortic valve replacement, 60 required mitral and aortic valve replacement, and eight also required associated tricuspid valve replacement (six mitral valve replacement; two mitral plus aortic valve replacement). Analysis of 34 preoperative and operative variables showed the treatment groups to be well randomized. In-hospital mortality was not significantly different among patients receiving the three prostheses for aortic valve replacement (7.6% overall) and mitral plus aortic valve replacement (10% overall), but there was a higher in-hospital mortality for patients undergoing mitral valve replacement with the Carpentier-Edwards prosthesis (15.5% compared with 8.8% overall; p = .03). This difference could not be explained on the basis of any preoperative or operative variable. Median follow-up was 5.6 (range 2.8 to 8.3) years. Actuarial survival after mitral valve replacement was 56.7 +/- 7.0% at 7 years, that after aortic valve replacement was 69.6 +/- 9.6% at 7 years, and that after mitral plus aortic valve replacement was 62.5 +/- 20.0% at 7 years. There was no significant difference in actuarial survival of patients receiving the three prostheses within the mitral, aortic, and mitral plus aortic valve replacement groups, nor was there a difference when these groups were amalgamated. Thirty-seven patients required reoperation for valve failure (15 with Björk-Shiley, 12 with Hancock, and 10 with Carpentier-Edwards valves; p = NS) and 11 died at reoperation (four with Björk-Shiley, four with Hancock, and three with Carpentier-Edwards valves; overall operative mortality 29.7%). Up to 7 years after surgery, there was no significant difference in the incidence of thromboembolism in patients with different prostheses undergoing mitral or aortic valve replacement. There were too few patients undergoing mitral plus aortic valve replacement for meaningful comparison. There was no significant beneficial effect of anticoagulants in patients undergoing mitral or aortic valve replacement with porcine prostheses, but patients were not randomly allocated to anticoagulant treatment. All patients with Björk-Shiley prostheses received anticoagulants.(ABSTRACT TRUNCATED AT 400 WORDS)