Forty years after the first implantation of caval filters, there is still no indication for implantation validated by a controlled clinical trial. This fact may be explained by our poor understanding of the evolution of thromboembolic disease, especially in certain groups of patients. The absolute contra-indications to heparin therapy would seem to be a logical indication for a caval filter. In cases of a relative contra-indication to anticoagulants, the physician has to rely on his clinical judgement and the decision will be taken case by case. In patients with suspected pulmonary embolism under anticoagulant therapy, it is also logical to check that anticoagulation is effective, and to request proof of embolism, to assess its risk and that due to thrombosis before considering a caval filter. The prophylactic implantation of a caval filter is a very controversial indication whether the thrombus is proximal in the ilio-caval region, extensive, not uncommon despite treatment, or floating. For groups said to be at high risk of thromboembolism (elderly, malignant disease or multiple injuries), there is no consensus because of the discordant results in the literature. The implantation of a filter would seem to be justified in patients with chronic cor pulmonale after pulmonary embolectomy. The value of a temporary caval filter during thrombolysis has not been demonstrated; there are hopes that temporary filters "of long duration" will provide filtration of the vena cava during vulnerable periods. The results of the first controlled trial (PREPIC) are eagerly awaited and should rationalise the indications of inferior vena cava filters.