Atrial fibrillation in the absence of rheumatic valvular disease is associated with a fivefold to sevenfold increased risk of ischemic stroke. The Stroke Prevention in Atrial Fibrillation Study, a multicenter, randomized trial, compared 325 mg/day aspirin (double-blind) or warfarin with placebo for prevention of ischemic stroke and systemic embolism (primary events), and included 1,330 inpatients and outpatients with constant or intermittent atrial fibrillation. During a mean follow-up of 1.3 years, the rate of primary events in patients assigned to placebo was 6.3% per year and was reduced by 42% in those assigned to aspirin (3.6% per year; p = 0.02; 95% confidence interval, 9-63%). In the subgroup of warfarin-eligible patients (most less than 76 years old), warfarin dose-adjusted to prolong prothrombin time to 1.3-fold to 1.8-fold that of control reduced the risk of primary events by 67% (warfarin versus placebo, 2.3% versus 7.4% per year; p = 0.01; 95% confidence interval, 27-85%). Primary events or death were reduced 58% (p = 0.01) by warfarin and 32% (p = 0.02) by aspirin. The risk of significant bleeding was 1.5%, 1.4%, and 1.6% per year in patients assigned to warfarin, aspirin, and placebo, respectively. Aspirin and warfarin are both effective in reducing ischemic stroke and systemic embolism in patients with atrial fibrillation. Because warfarin-eligible patients composed a subset of all aspirin-eligible patients, the magnitude of reduction in events by warfarin versus aspirin cannot be compared. Too few events occurred in warfarin-eligible patients to directly assess the relative benefit of aspirin compared with warfarin, and the trial is continuing to address this issue. Patients with nonrheumatic atrial fibrillation who can safely take either aspirin or warfarin should receive prophylactic antithrombotic therapy to reduce the risk of stroke.