BACKGROUND Many centers advocate bioprosthetic valves in the elderly to avoid anticoagulation, in particular when patient survival is less than the expected valve durability. Because expected survival in the elderly is increasing and age-specific risk of anticoagulation in the elderly is not known, we examined valve- and anticoagulation-related morbidity in elderly patients after aortic valve replacement (AVR) with bioprostheses or mechanical prostheses. RESULTS Between January 1980 and June 1994, 211 patients age > or = 70 years underwent isolated AVR; there were 109 men (52%) and 102 women (48%). Mean age was 75.9 +/- 4.8 years. Aortic stenosis was present in 194 (92%) patients. Bioprostheses were used in 145 (69%) and mechanical prostheses were used in 66 (31%). Chronic anticoagulation was maintained in all patients with a mechanical valve and in 18 patients (12%) with a bioprosthetic valve. Follow-up data were obtained for 98% (194 of 197) of hospital survivors at a mean follow-up of 3.8 years. Operative mortality was 6.6%; survival at 3 and 5 years was 75.3 +/- 3% and 64.6 +/- 4%, respectively. There was no significant difference in operative or late mortality between patient groups. Rates of freedom from thromboembolic events, endocarditis and anticoagulant-related hemorrhage for bioprosthetic and mechanical valve patients were similar. Prosthetic failure was identified in three bioprosthetic valves (2%); furthermore, the 4 patients in the series who required reoperation had received bioprostheses at the first operation. CONCLUSIONS In conclusion, (1) elderly patients undergoing isolated AVR can be managed with either mechanical or bioprosthetic valves with similar early and late risk, as long as there are no specific contraindications to anticoagulation; (2) anticoagulation-related risk of hemorrhage is low in this group of elderly patients; and (3) the low but significant risk of reoperation following the use of bioprostheses suggests that mechanical valves may be underused in the elderly.