Retrospective follow-up data on 458 consecutive patients who received a Hancock, Carpentier-Edwards (C-E) or Ionescu-Shiley (I-S) bioprosthesis in the aortic valve position between April 1978 and December 1981 are reviewed. A total of 461 valves (184 Hancock, 131 C-E and 146 I-S) were available for study of the incidence of primary tissue valve failure after 5 to 9 years of follow-up. Cumulative follow-up was 1,016 patient-years for patients with Hancock valve, 688 for the C-E and 767 for the I-S group. Of the 397 prostheses at risk (154 Hancock, 120 C-E and 123 I-S), 36 instances of primary tissue valve failure occurred (12 Hancock, 7 C-E and 17 I-S). On an actuarial basis, the calculated probability of freedom from primary tissue valve failure was 88 +/- 4% for the Hancock group, 87 +/- 6% for the C-E and 51 +/- 17% for the I-S at 9 years. The linear incidence of tissue valve failure was 1.2 failing valves per 100 patient-years for the Hancock group, 1 for the C-E and 2.2 for the I-S. In a cohort of patients older than 40 years of age at the time of valve replacement, the rate of freedom from primary failure was 98 +/- 1%, 87 +/- 9% and 44 +/- 22% for the Hancock, C-E and I-S groups, respectively, at 9 years. Comparison of actuarial curves disclosed a meaningful difference between the pericardial valve and the Hancock and C-E porcine bioprostheses at 9-year follow-up.(ABSTRACT TRUNCATED AT 250 WORDS)