Determination of the optimal duration of anticoagulant treatment for venous thromboembolic disease (VTED) is a major step in the management of patients with this disease. The assessment depends on the identification of two sets of risk factors: those for recurrence after anticoagulant treatment is stopped and those for hemorrhage in cases of prolonged treatment. Nonetheless, the determination of the optimal duration remains controversial. Recent data finally make it possible to clarify this decision. Recent treatment trials demonstrate that patients at high risk of recurrence receive no sustained benefit from a prolonged but limited anticoagulant treatment. In other words, the choice is simplified: either the risk is low, and treatment for 3months is sufficient, or the risk is high, and treatment must be envisioned for an unlimited duration. Adequate identification of patients eligible for short or unlimited treatment is more crucial than ever and depends on the presence of determinant clinical variables, as the information from laboratory or morphologic tests is generally marginal. The risk of thromboembolic recurrence is low when the initial episode is triggered by a major reversible factor, and a short treatment of 3months is thus indicated. These inducing factors are mainly surgery, lower limb injuries, immobilization for a medical condition, pregnancy, or use of combined estrogen-progestin contraceptives. Among patients with VTED not induced by these factors, the risk of recurrence is high and requires planning anticoagulant treatment for an unlimited duration. Nonetheless, the risk of hemorrhage is a major constraint to such unlimited treatment. Accordingly, the perspectives for secondary prevention that is equally effective but has a lower risk of hemorrhage are currently under evaluation. Finally, patients with cancer are in a separate category, with a very high risk of recurrence that justifies treatment for at least 6months.