Oral anticoagulants (OA) are the drug of choice for stroke prevention in patients with non-rheumatic atrial fibrillation (NRAF). This clear benefit/risk ratio comes from several randomized clinical trials (RCT) in which highly selected patients were strictly monitored. The aim of this study was to ascertain whether the safety of OA was also obtained outside the setting of clinical trials in consecutive patients starting treatment and routinely followed at Italian anticoagulation clinics. A total of 433 patients with NRAF were enrolled in the ISCOAT study and followed up for a mean of 1.4 years. Two patients (0.3% per year) suffered from a complete non-fatal ischemic stroke, 8 patients (1.3% per year) died of thrombosis-related vascular death, and 11 patients (11 events, 1.8% per year) suffered from major bleedings (2 fatal). Major bleeding occurred more frequently in patients >75 years of age (6 events, 5.1% per year) than in younger patients (5 events, 1.0% per year). The cumulative incidence of major bleeding in patients over 75 years of age (10.8%; 95% CI, 1.8-19.8) was significantly higher than in younger patients (2.8%; 95% CI, 0.3-5.3, p = 0.006). Major primary bleeding unrelated to organic lesions (7 patients, 1 male and 6 females) occurred in 5 elderly patients (>75 years old) with a cumulative incidence (9.6%; 95% CI 0.8-18.4) significantly higher than in younger patients (1.2%; 95% CI, 0-3.0, p = 0.0003). Univariate analysis revealed a higher frequency of major primary bleeding in females, in diabetic patients and in in those who had suffered a previous thromboembolic event. Multivariate analysis revealed that only age grater than 75 years was independently related to major primary bleedings (RR 6.6; 95% CI 1.2-37, p = 0.032). Minor bleedings (n = 27) were not more frequent in elderly patients (6% vs 4% per year, p = ns). Patients were kept at optimal intensity of treatment for 63% of the time. These data confirm the efficacy of OA but identify elderly patients as a high risk group of major bleeding.