The Canadian Atrial Fibrillation Anticoagulation Study was a randomized double-blind placebo-controlled trial to assess the potential of warfarin to reduce systemic thromboembolism and its inherent risk of hemorrhage. As a result of the publication of two other "positive" studies of similar design and objective, this study was stopped early before completion of its planned recruitment of 630 patients. There were 187 patients randomized to warfarin and 191 to placebo. Permanent discontinuation of study medication occurred in 26% of warfarin-treated and 23% of placebo-treated patients. The target range of the international normalized ratio was 2 to 3. For the warfarin-treated patients, the international normalized ratio was in the target range 43.7% of the study days, above it 16.6% of the study days and below it 39.6% of the study days. Fatal or major bleeding occurred at annual rates of 2.5% in warfarin-treated and 0.5% in placebo-treated patients. Minor bleeding occurred in 16% of patients receiving warfarin and 9% receiving placebo. The primary outcome event cluster was nonlacunar stroke, noncentral nervous systemic embolism and fatal or intracranial hemorrhage. Events were included in the primary analysis of efficacy if they occurred within 28 days of permanent discontinuation of the study medication. The annual rates of the primary outcome event cluster were 3.5% in warfarin-treated and 5.2% in placebo-treated patients, with a relative risk reduction of 37% (95% confidence limits, -63.5%, 75.5%, p = 0.17).(ABSTRACT TRUNCATED AT 250 WORDS)