From November, 1985, to June, 1988, 1007 outpatients with chronic non-rheumatic atrial fibrillation (AF) entered a randomised trial; 335 received anticoagulation with warfarin openly, and in a double-blind study 336 received aspirin 75 mg once daily and 336 placebo. Each patient was followed up for 2 years or until termination of the trial. The primary endpoint was a thromboembolic complication (stroke, transient cerebral ischaemic attack, or embolic complications to the viscera and extremities). The secondary endpoint was death. The incidence of thromboembolic complications and vascular mortality were significantly lower in the warfarin group than in the aspirin and placebo groups, which did not differ significantly. 5 patients on warfarin had thromboembolic complications compared with 20 patients on aspirin and 21 on placebo. 21 patients on warfarin were withdrawn because of non-fatal bleeding complications compared with 2 on aspirin and none on placebo. Thus, anticoagulation therapy with warfarin can be recommended to prevent thromboembolic complications in patients with chronic non-rheumatic AF.