BACKGROUND: Anticoagulants are more effective than antiplatelet agents at reducing stroke risk in patients with atrial fibrillation, but whether this benefit outweighs the increased risk of bleeding in elderly patients is unknown. We assessed whether warfarin reduced risk of major stroke, arterial embolism, or other intracranial haemorrhage compared with aspirin in elderly patients. METHODS: 973 patients aged 75 years or over (mean age 81.5 years, SD 4.2) with atrial fibrillation were recruited from primary care and randomly assigned to warfarin (target international normalised ratio 2-3) or aspirin (75 mg per day). Follow-up was for a mean of 2.7 years (SD 1.2). The primary endpoint was fatal or disabling stroke (ischaemic or haemorrhagic), intracranial haemorrhage, or clinically significant arterial embolism. Analysis was by intention to treat. This study is registered as an International Standard Randomised Controlled Trial, number ISRCTN89345269. FINDINGS: There were 24 primary events (21 strokes, two other intracranial haemorrhages, and one systemic embolus) in people assigned to warfarin and 48 primary events (44 strokes, one other intracranial haemorrhage, and three systemic emboli) in people assigned to aspirin (yearly risk 1.8%vs 3.8%, relative risk 0.48, 95% CI 0.28-0.80, p=0.003; absolute yearly risk reduction 2%, 95% CI 0.7-3.2). Yearly risk of extracranial haemorrhage was 1.4% (warfarin) versus 1.6% (aspirin) (relative risk 0.87, 0.43-1.73; absolute risk reduction 0.2%, -0.7 to 1.2). INTERPRETATION: These data support the use of anticoagulation therapy for people aged over 75 who have atrial fibrillation, unless there are contraindications or the patient decides that the benefits are not worth the inconvenience.