Transvenous interruption of the inferior vena cava was successful in 45 of 47 consecutive patients treated during a 41/2-year period. Greenfield filters were used in 40, and Mobin-Uddin umbrellas in 5 others. Technical details of the insertion procedure are of particular importance for successful implantation of the Greenfield device. Follow-up data were available for 34 of 36 surviving patients and autopsy findings for 4 of the 11 who had died. Patency of the vena cava was assessed by radionuclide scan, venography, or autopsy. Of 33 patients with Greenfield filters who were evaluated, only one had an occluded filter, for a patency rate of 97%. One of the three patients with a Mobin-Uddin umbrella had caval thrombosis. Severe postoperative venous stasis was seen in one patient with a Greenfield filter and one with a Mobin-Uddin umbrella. Significant distal migration or angulation of Greenfield filters was observed in six patients and is a theoretical cause for recurrent embolization. The 30-day mortality rate for the patients who had transvenous procedures was 13%, but no deaths were attributed to the procedure itself. There were no clinically evident pulmonary emboli during the follow-up period. In the present study we have documented low operative morbidity and mortality rates as well as complete protection from recurrent pulmonary embolism using transvenous interruption of the inferior vena cava. Although the Greenfield filter is somewhat bulkier and more difficult to insert than the Mobin-Uddin umbrella, it has a clear advantage because of its superior long-term patency.