Various predictive scores for vitamin K antagonist (VKA)-related bleeding have been developed and validated in outpatients and in patients treated for specific indications (when VKAs are used under optimal therapeutic conditions). However, there are few published data on the evaluation of bleeding risk factors in hospitalized, at-risk patients (with a high international normalized ratio [INR]) treated with VKAs. The objective of the present study was to identify the most relevant bleeding risk factors in 906 VKA-treated patients with an INR of 5 or more hospitalized in a French university medical center.Over a 2-year period, we screened all consecutive VKA-treated adults with a risk of major bleeding (defined as an INR ≥ 5 on admission). Demographic and clinical characteristics, medications, and bleeding characteristics were recorded prospectively.The overall incidence of bleeding was 26.6% (serious bleeding: 21.4%; fatal bleeding: 5.4%). An INR ≥ 8.5, a history of recent digestive tract lesions, trauma in the preceding 2 weeks, and known noncompliance were independent risk factors for bleeding and serious bleeding.Our present findings emphasize that VKAs should not be prescribed to patients with a high risk of bleeding (noncompliant patients and those with recent trauma or recent gastrointestinal lesions). It is essential to monitor the INR on a frequent basis and adjust oral anticoagulant treatment appropriately.