Isolated aortic (n = 107), mitral (n = 63), and tricuspid (n = 1) valve replacement and 28 double-valve replacements were performed with a second generation of pericardial valves, the Mitroflow valve, in 199 patients from March 1983 to December 1986. Follow-up (total, 1,058 patient-years) was extended to 106 months and 91.5% complete. Mean age was 58 +/- 13 years. The operative mortality included 22 deaths, non-cardiac-related in 7. The actuarial probability of survival for all patients was 66% +/- 4% at 8.5 years. There were no significant differences between patients with aortic valve replacement, mitral valve replacement, or double-valve replacement. The rate of thromboembolic events, antithromboembolic therapy-related hemorrhage, periprosthetic leak, and endocarditis is extremely minimal. Structural valve dysfunction occurred at a rate of 3.2% +/- 0.5%/patient-year. Actuarial freedom from the event was 94.6% +/- 1.7% at 5 years and 63.7% +/- 6.5% at 8.5 years for all valves. There were no difference in structural valve dysfunction rate between patients having aortic, mitral, or double-valve replacement. Thirty-five patients were reoperated on (3.4 +/- 0.6%/patient-year for all). The rate of all valve-related morbidity and mortality was 5.6% +/- 0.7%/patient-year for all patients, actuarial freedom from the event being 44% +/- 7% at 8.5 years. These data suggest that the excellent hemodynamic characteristics of the valve are balanced by a risk of valve failure that is slightly increased when compared with porcine valves.