Hematologists are often asked to treat patients with venous thromboembolic disease. Although anticoagulation remains the primary therapy for venous thromboembolism, vena caval filters are an important alternative when anticoagulants are contraindicated. To assess the evidence supporting the utility of these devices, a comprehensive review of the English language literature was performed. Except for one randomized trial, the vena caval filter literature consists of case series or consecutive case series. The mean duration of follow-up for each of the 5 filter types varies from 6 to 18 months. All are about equally effective in the prevention of pulmonary embolism (2.6%-3.8%). Deep venous thrombosis (6%-32%) and inferior vena cava thrombosis (3.6%-11.2%) after filter placement vary widely among different filter types primarily because of differences in outcome assessment. Thrombosis at the insertion site is a common complication of filter placement (23%-36%). In view of the absence of randomized comparisons, no filter can be designated as superior in safety or efficacy. Vena caval filters represent a potentially important but poorly evaluated therapeutic modality in the prevention of pulmonary emboli. Randomized trials are necessary to establish the appropriate place for vena caval filters in the treatment of venous thromboembolic disease. (Blood. 2000;95:3669-3677)