We read the recent report of Qureshi et al reporting the outcomes of restarting anticoagulation after gastrointestinal bleeding (GIB) in patients with atrial fibrillation. The investigators found that patients who resumed taking warfarin in the first week were at a higher risk of recurrent GIB (incidence ratio of 19.3%), whereas a lower risk of thromboembolism and mortality was associated with this early reinitiation of warfarin. Because their findings are important to both current practice and future research, several limitations of this study deserve attention. Nearly 40% of patients with warfarin restarted presented an upper GIB. Recommendations in upper GIB emphasized early risk stratification, using validated prognostic scales, and early endoscopy (within 24 hours). So, it would be very important to know the mean Rockall risk score, the best score to estimate the risk of both rebleeding and mortality, to better understand the high rate of recurrent bleeding observed in this study. In the same way, the 3 most common approaches for reversal of warfarin are (1) discontinuation of oral anticoagulant therapy, (2) oral or intravenous administration of agents that promote coagulation, and (3) direct reconstitution of the patient’s coagulation factors through intravenous administration. Although all patients presented with a major GIB on admission, only <48.5% and <28% of patients received fresh frozen plasma and vitamin K, respectively. So, the possible absence of “coagulation normalization” could also explain the high rate of hemorrhagic recurrence. Finally, stratifying patients according to their risk of thromboembolism is based on patients’ clinical indication for antithrombotic therapy and the presence of co-morbidities. For perioperative arterial and thromboembolism risk, a moderate risk is suggested for patients with a CHADS ₂ score of 3 and a low risk is suggested in patients with a CHADS ₂ score of 0 to 2. So, it appears pertinent to propose anticoagulant treatment withdrawal for as short a period as possible in only situations involving a high risk of thrombosis, as here the risk of thromboembolism could be higher than the risk of hemorrhage. It is the aim of our ongoing clinical trial to evaluate the risk and/or benefit of early versus late resumption of anticoagulation in patients with major non–trauma-related hemorrhage occurring while receiving anticoagulant treatment for a high risk of thrombosis (NCT02091479).