Myocardial infarction survivors have an increased risk of stroke, which is reduced with long-term anticoagulant therapy. However, an estimated 10-times increase in risk of bleeding during such treatment has been reported. We evaluated the risk of stroke in patients after a myocardial infarction and examined the relationship of the risk of intracranial hemorrhage or cerebral infarction and the intensity of anticoagulant therapy. The study population consisted of 3,404 post-myocardial infarction patients who took part in a randomized, double-blind, placebo-controlled trial. Patients were randomized to treatment with anticoagulants (international normalized ratio range, 2.8-4.8) or matching placebo. Mean follow-up was more than 3 years. The incidence of stroke analyzed on "intention-to-treat'' was 0.7 per 100 patient-years in the anticoagulant patients against 1.2 in placebo, a hazard ratio of 0.60, with 95% confidence interval of 0.40 to 0.90. In the anticoagulation group, 15 patients had cerebral infarction and 17 an intracranial bleeding, 3 of which occurred after withdrawal of treatment. In the placebo group, the numbers were 43 and 2. Of the 14 intracranial bleeds during anticoagulation, 6 occurred at an international normalized ratio between 3.0 and 4.0 and 8 at greater than 4.0. These results confirm that long-term anticoagulant therapy substantially reduces the risk of stroke in postmyocardial infarction patients. The increased risk of bleeding complications associated with anticoagulant therapy is offset by a marked reduction in ischemic events. The risk of intracranial bleeding is directly related to the intensity of anticoagulant treatment.