There is controversy over whether elderly patients benefit from the durability of mechanical valves when balanced against the risk of anticoagulation. From 1976 to 1993, 576 patients 65 years old or older underwent isolated valve replacement with mechanical (n = 250) or bioprosthetic valves (n = 326). Total follow-up was 2,222 patient-years. Probability of survival and freedom from thromboembolism and prosthetic valve endocarditis were not different between the two groups. There was a significant difference (p = 0.015) in freedom from anticoagulant-related hemorrhage. Two patients with mechanical prostheses and 7 patients with bioprostheses were reoperated. However, actuarial freedom from reoperation was not different (p = 0.73) in both groups, with no hospital mortality, whereas mortality from thromboembolic events and anticoagulant-related hemorrhage was three times higher in patients with mechanical prostheses as compared with patients with bioprostheses (1.08% versus 0.36% per patient-year). The benefit from the durability of mechanical valves, compared with bioprostheses, is smaller than expected because of the limited number of patients exposed to the onset of bioprosthetic structural deterioration. Elderly patients without absolute indication for anticoagulation should preferentially receive bioprostheses for valvular replacement.