From November, 1973, through June, 1978, 428 operations in 425 patients were performed for replacement of aortic, mitral, or aortic plus mitral valves, utilizing 277 Hancock and 180 Carpentier-Edwards bioprostheses. Actuarially determined survival at 36 months was similar for all three groups and compared favorably with our experience with the Björk-Shiley prosthesis. Certain patient-related variables influencing late survival were identified by multivariate analysis and included previous operation for congenital heart disease, coronary artery bypass grafting in nonaortic valve replacement, race (black), age at operation, and New York Heart Association Functional Class. A small but definite incidence of thromboembolism occurred in all three groups, again similar to our experience with the Björk-Shiley prosthesis. Multivariate analysis identified four factors influencing risk of thromboembolism: previous cardiac operation, age, double-valve replacement, and rhythm at discharge. Valve degeneraation occurred, primarily in children and young adults. Over the medium term, the porcine bioprosthesis compared favorably with mechanical prostheses in terms of survival, function, and thromboembolism. Certain patient-related variables affecting survival may be modified by earlier surgical intervention.