In this issue of CHEST (see page 1390), Shireman et al20 propose a new contemporary bleeding risk model for elderly patients receiving warfarin, which tries to fully address the difficulties and complexities involved in anticoagulation administration. Their model, which was validated in > 26,000 patients, incorporates eight variables to determine the bleeding risk: age ≥ 70 years, gender, remote bleed, recent bleed, alcohol/drug abuse, diabetes, anemia, and antiplatelet therapy use. They demonstrate bleeding rates of 0.9%, 2.0%, and 5.4% for the low-, moderate-, and high-risk groups, respectively. This study uses a sizeable population, which adds weight and increases the power of the study, thereby validating its use. Clinical trials have been accused of simulating perfect conditions and often involve exclusion of older populations, but in the article by Shireman et al,20 43% of the sample size was > 80 years old and this is certainly more representative of the AF population. Risk factors such as concomitant alcohol use, drug abuse, and gender were taken into account. Differentiation was also made between recent and remote bleeds, and lifetime incidence rather than recent myocardial infarction was considered as a variable. The model by Shireman et al20 was also compared with the OBRI and Kuijer models, with good differentiation between low-, moderate-, and high-risk groups, whereas the Kuijer model was unable to show any difference between the intermediate-risk and high-risk groups. However, an eight-variable risk stratification schema is obviously cumbersome and may limit its applicability. An obvious risk factor that has not been taken into account is the range (or variability) of international normalized ratio (INR) values. We do know that INR values > 3.0 are associated with a greater risk of bleeding and that the risk increases with further increases in the INR. Their study is also limited to a 90-day follow-up period and although a sizeable proportion of warfarin associated bleeds occurred within 90 days, a longer-term follow-up would be desirable to establish a much clearer picture.