To examine indications for, and morbidity and mortality rate of, inferior vena cava filter insertion at a community hospital, the records of 69 patients who received inferior vena cava filters were reviewed. Patients were assigned to three groups--group 1, 45 patients with pulmonary embolism or deep venous thrombosis and a contraindication to anticoagulation; group 2, 14 patients with a diagnosis as in group 1, who received filters without consideration to anticoagulation, and group 3, ten patients with clinically suspected deep venous thrombosis and no objective assessment of the process. Indications for filter placement were recorded. Morbidity and in-hospital mortality rates were 29 and 49 percent, 43 and 36 percent and 10 and 30 percent for groups 1, 2 and 3, respectively (29 and 43 percent overall). Only patients in group 1 had documented indications for caval interruption. Results compared unfavorably with complication and mortality rates reported previously. Non-selective use of inferior vena cava filters is associated with unacceptable morbidity and mortality rates. Strict indications for filters must be well documented.