Mitral stenosis (MS) in association with atrial fibrillation (AF) is a clinical condition at high risk for systemic thromboembolism. Although oral anticoagulants greatly reduce the incidence of thromboembolism in these patients, the optimal intensity of treatment has never been tested in specific clinical trials, and current recommendations are derived from studies of nonrheumatic AF. In this study we tested the effectiveness of two different intensities. The study design was carried out as an open randomized prospective study in an anticoagulation clinic. We randomized 103 patients with MS and AF to a low (target INR = 2) or moderate (target INR = 3) anticoagulation regimen. The primary end points were systemic thromboembolism, major bleeding and vascular death. During a mean follow-up of 4.5 years, 1 systemic embolism occurred in the low intensity group (0.41 per 100 pt/yrs, CI 0.01-2.3), and 1 minor stroke occurred in the moderate intensity group (0.40 per 100 pt/yrs, CI 0.01-2.3; p = ns). Major bleeding occurred in 8 patients, with 3 in the low intensity (1.25 per 100 pt/yrs) and 5 in the moderate intensity group (2.0 per 100 pt/yrs, Incidence Rate Ratio 0.6, CI 0.1-3.1; p = ns). Total events (systemic embolism, major bleeding and vascular death) occurred in 7 low intensity patients and 8 moderate intensity patients. As expected, minor bleeding was more frequent in the moderate intensity group of patients, who actually had more intense treatment and required closer monitoring of oral anticoagulant treatment. These data suggest that low intensity anticoagulation, as performed in an anticoagulation clinic, is effective and safe in high risk patients with MS and AF.