SummaryBackground. The long-term administration of oral anticoagulants to patients with mechanical heart valve prostheses is generally accepted. However, the appropriate intensity of oral anticoagulant treatment in these patients is still controversial.Methods and Results. From March 1991 to March 1994, patients referred to the Padova Thrombosis Center who had undergone mechanical heart valve substitution at least 6 months earlier were randomly assigned to receive oral anticoagulants at moderate intensity (target INR = 3) or moderate-high intensity (target INR = 4). Principal end points were major bleeding, thromboembolism and vascular death. Minor bleeding was a secondary end-point.A total of 104 patients were assigned to the target 3 group and 101 to the target 4 group; they were followed for from 1.5 years to up 4.5 years (mean, 3 years). Principal end-points occurred in 13 patients in the target 3 group (4 per 100 patient-years) and in 20 patients in the target 4 group (6.9 per 100 patient-years). Major hemorrhagic events occurred in 15 patients, 4 in the target 3 group (1.2 per 100 patient-years) and 11 in the target 4 group (3.8 per 100 patient-years) (p = 0.019). The 12 recorded episodes of thromboembolism, 4 of which consisted of a visual deficit, were all transient ischemic attacks, 6 in the target 3 group (1.8 per 100 patient-years) and 6 in the target 4 group (2.1 per 100 patient- years). There were 3 vascular deaths in each group (0.9 and 1 per 100 patient-years for target 3 and target 4 groups, respectively). Minor bleeding episodes occurred 85 times (26 per 100 patient-years) in the target 3 group and 123 times (43 per 100 patient-years) in the target 4 group (p = 0.001).Conclusions. Mechanical heart valve patients on anticoagulant treatment who had been operated on at least 6 months earlier experienced fewer bleeding complications when maintained on a moderate intensity regimen (target INR = 3) than those on a moderate-high intensity regimen (target INR = 4). The number of thromboembolic events and vascular deaths did not differ between the two groups.