Various methods have been described to evaluate efficacy of anticoagulant therapy using the international normalized ratio (INR). We compared the following approaches: (1) total INR's or the most recent measurement; (2) percent time within therapeutic range, with INR changing directly or halfway between visits; and (3) total observation time assuming INR changing linearly. The study population comprised 1700 post myocardial infarction patients. Treatment comprised 3725 patient-years. There were 61,471 INR assessments with target therapeutic level of 2.8-4.8. Acenocoumarol as well as phenprocoumon were employed. Therapeutic achievement in the first months of treatment was low: less than 60% of INR's were in range. Treatment stabilized after 6 months. Patients on acenocoumarol were within range 70% of the time compared to 80% for phenprocoumon. Method 3 is preferred because it incorporates time and is capable of calculating incidence rates at different INR levels. Our findings call for an urgent improvement of standard of anticoagulant control in the first months following commencement of treatment.