Based on the perception of an unacceptably high risk of major bleeding in many older persons receiving anticoagulant therapy, Dr Pechlaner concludes that many older patients with atrial fibrillation might not derive overall benefit for anticoagulant therapy. A number of references used to support his conclusions are open to critique in the context of today's clinical practice. For example, the study by Beyth et al1 was performed in the 1980s, a time when higher-intensity anticoagulation was the norm. In fact, the mean international normalized ratios of persons in this study were greater than 3.0, a level clearly shown to increase bleeding risk without providing incremental stroke protection benefit. Therefore, the estimates of bleeding risk from this study are likely higher than that found in current practice, with the recommended target international normalized ratio being 2.5 (range, 2.0-3.0), and the clinical usefulness of their bleeding risk stratification scheme is open to question. Also, in the cost-effectiveness study of the value of anticoagulant therapy in the oldest old (up to 100 years of age) by Desbiens,2 a few questionable assumptions about the risk of bleeding (especially hemorrhagic stroke) in the oldest age groups were made. Desbiens based the extrapolation of intracranial bleeding risk in older persons on data from the Stroke Prevention in Atrial Fibrillation II study.3 The anticoagulation intensity (international normalized ratio, 2.0-4.5) of this study was also higher than what is used presently, leading to a bleeding rate higher than that found in other atrial fibrillation studies.4 Therefore, the extrapolated estimates by Desbiens of intracranial bleeding risk in older age groups taking warfarin (including a 10% yearly risk in persons 100 years old) appear to be a serious overestimation. Furthermore, the robustness of his analysis is compromised when sensitivity analysis showed that the model was quite sensitive to variations in this parameter. Finally, while these studies suggest an increase in bleeding risk as one ages, other studies5- 6 have not found this phenomenon, though they too have some methodological difficulties.