Atrial fibrillation is associated with an increased risk of ischemic stroke. Data on individual patients were pooled from five recently completed randomized trials comparing warfarin (all studies) or aspirin (the Atrial Fibrillation, Aspirin, Anticoagulation Study and the Stroke Prevention in Atrial Fibrillation Study) with control in patients with atrial fibrillation. The purpose of the analysis was to (1) identify patient features predictive of a high or low risk of stroke, (2) assess the efficacy of antithrombotic therapy in major patient subgroups (eg, women), and (3) obtain the most precise estimate of the efficacy and risks of antithrombotic therapy in atrial fibrillation. For the warfarin-control comparison there were 1889 patient-years receiving warfarin and 1802 in the control group. For the aspirin-placebo comparison there were 1132 patient-years receiving aspirin and 1133 receiving placebo. The daily dose of aspirin was 75 mg in the Atrial Fibrillation, Aspirin, Anticoagulation Study and 325 mg in the Stroke Prevention in Atrial Fibrillation Study. To monitor warfarin dosage, three studies used prothrombin time ratios and two used international normalized ratios. The lowest target intensity was a prothrombin time ratio of 1.2 to 1.5 and the highest target intensity was an international normalized ratio of 2.8 to 4.2. The primary end points were ischemic stroke and major hemorrhage, as assessed by each study. At the time of randomization the mean age was 69 years and the mean blood pressure was 142/82 mm Hg. Forty-six percent of the patients had a history of hypertension, 6% had a previous transient ischemic attack or stroke, and 14% had diabetes. Risk factors that predicted stroke on multivariate analyses in control patients were increasing age, history of hypertension, previous transient ischemic attack or stroke, and diabetes. Patients younger than 65 years who had none of the other predictive factors (15% of all patients) had an annual rate of stroke of 1.0%, 95% confidence interval (CI) 0.3% to 3.0%. The annual rate of stroke was 4.5% for the control group and 1.4% for the warfarin group (risk reduction, 68%; 95% CI, 50% to 79%). The efficacy of warfarin was consistent across all studies and subgroups of patients. In women, warfarin decreased the risk of stroke by 84% (95% CI, 55% to 95%) compared with 60% (95% CI, 35% to 76%) in men. The efficacy of aspirin was not as consistent. The risk reduction with 75 mg of aspirin in the Atrial Fibrillation, Aspirin, Anticoagulation Study was 18% (95% CI, 60% to 58%), and with 325 mg of aspirin in the Stroke Prevention in Atrial Fibrillation Study the risk reduction was 44% (95% CI, 7% to 66%). When both studies were combined the risk reduction was 36% (95% CI, 4% to 57%). The annual rate of major hemorrhage (intracranial bleeding or a bleed requiring hospitalization or 2 units of blood) was 1.0% for the control group, 1.0% for the aspirin group, and 1.3% for the warfarin group. In these five randomized trials warfarin consistently decreased the risk of stroke in patients with atrial fibrillation (a 68% reduction in risk) with virtually no increase in the frequency of major bleeding. Patients with atrial fibrillation younger than 65 years without a history of hypertension, previous stroke or transient ischemic attack, or diabetes were at very low risk of stroke even when not treated. The efficacy of aspirin was less consistent. Further studies are needed to clarify the role of aspirin in atrial fibrillation.