Computer-assisted oral anticoagulant dosage is being increasingly used to meet growing demands for oral anticoagulation. The DAWN AC is one of the most widely used computer-dosage programs. Evidence of its value and that of other computer programs has been based previously only on laboratory evidence of "time in target INR range" (TIR) not on clinical safety in practice. A five-year international randomised clinical study of computer assistance with the DAWN AC program compared with manual dosage in 2,631 patients has been performed at 13 centres with established expertise in oral anticoagulation mainly in the EU. Safety assessment have been based on the comparison of bleeding or thrombotic events with DAWN AC compared with manual dosage in a randomised study. Safety of the DAWN AC program has been demonstrated. Clinical events of bleeding and thrombosis were almost identical with the experienced manual dosage group. Therapeutic control improved with DAWN AC to 66.8% from 63.4% TIR. The program failed to provide a dosage recommendation on only 5.7% of occasions. At a group of experienced centres with a special interest in oral anticoagulation, the DAWN AC computer-dosage program proved as safe clinically as manual dosage by experienced medical staff. With DAWN AC, laboratory control was improved, the difference being highly significant. The results should reassure hospitals and community clinics that the DAWN AC program is safe and facilitate greater and longer provision of warfarin treatment where required.