To ascertain relations among site, incidence, and mechanisms of clinically evident failure of porcine bioprosthetic heart valves, the frequency of failure of 1,110 valves in 1,001 adult operative survivors from January 1972 to January 1982 was reviewed and correlated with the pathologic features of 22 consecutive dysfunctional valves. There were 373 mitral, 519 aortic, and 109 double replacements, yielding for study 482 mitral and 628 aortic valves at risk. Infective endocarditis occurred in 1.9% (8 mitral, 7 aortic, and 4 double). Twenty-three valves (13 mitral and 10 aortic) with documented primary dysfunction were explanted, a mean of 55 months (range 9 to 94) after surgery. The primary dysfunction rate for the 333 valves implanted for greater than or equal to 5 years was 6.8% (11 of 161) for mitral and 4.1% (7 of 172) for aortic valves. Valves implanted for less than 5 years had a failure rate of 0.7%. The actuarially determined freedom from primary valve failure was 98 +/- 1% for mitral and 98 +/- 1% for aortic valves at 5 years and 79 +/- 7% for mitral and 91 +/- 4% for aortic valves at 10 years. Recovered valves (12 mitral and 10 aortic) with detailed morphologic analysis were functioning for a mean duration of 52 months (range 12 to 87). Causes of failure included calcification-related tears in 7 (4 mitral and 3 aortic, mean 66 months), tear without calcium deposits in 4 (4 mitral, mean 44 months), cuspal stiffening without tear but with calcium deposits in 2 (1 mitral and 1 aortic, mean 80 months) and thrombosis in 1 (aortic). Late primary dysfunction was most frequently a result of degenerative processes, especially calcification, often with secondary tears, but cuspal tears in the absence of calcium deposits and thrombosis predominated at shorter intervals.