The function of vena cava filters, preventing pulmonary embolism while maintaining caval patency, is associated with the design. Several characteristics have been reported. This report evaluates retrievable filter designs in comparison with previously marketed designs with respect to efficacy and safety. Three inferior vena cava (IVC) filters (Gunther Tulip, Bard Recovery, and the Cordis OptEase) were compared on the basis of design characteristics associated with function, shape, number of trapping levels, and fixation. Adverse events reported in the literature and to the US Food and Drug Administration Manufacturers and User Facility Device Experience Database (MAUDE) were summarized. The major differences among device types include the fixation, the volume and number of trapping levels, and the amount of metal in the IVC. The MAUDE registry reported adverse events that had been hypothesized from the analysis of in vitro and in vivo testing. The Recovery and OptEase filters had the highest number of clinically important reports. From 12 to 57% were retrieved between 3 days and 11 months. Adaptations made to facilitate retrieval led to unacceptable sequelae. The small number removed, the length of time they are left in place, and the risks associated with retrieval suggest that optional filters may not be equivalent to approved permanent devices.