Retrievable vena cava filters (R-VCF) are a recent addition to the therapeutic armamentarium for the prevention of pulmonary embolism. However, unlike permanent vena cava filters (P-VCF), outcomes data are limited regarding complication rates.This was a retrospective comparative analysis of consecutive patients undergoing placement of R-VCF vs P-VCF at Wake Forest University School of Medicine from January 2000 to December 2004. Data collected included demographics, procedural specifics, filter type, indications, and complications. Summary data are expressed as number (percentage) or mean +/- SD. Continuous and categorical variables were analyzed by using t and Fisher exact testing, as appropriate. Four additional patients with vena cava thrombosis were also referred to our institution for treatment during the study period, all with opposed biconical VCFs (OptEase and TrapEase filters) recently placed at other facilities. This last group of patients is described but not included in the analysis.A total of 189 VCF (165 P-VCF and 24 R-VCF) cases were examined. No significant differences in VCF groups were observed according to age, documented hypercoagulability, or concomitant anticoagulation. Significant differences were observed according to sex (30.3% of P-VCF vs 62.5% of R-VCF patients were female), morbid obesity (4.2% of P-VCF vs 25% of R-VCF patients), active malignancy (20% of P-VCF vs 41.7% of R-VCF patients), and indication for VCF placement. Over a median follow-up of 8.5 months, no case of significant hemorrhage, no VCF migration, and four cases of vena cava thrombosis were observed. Vena cava thrombosis was observed more frequently in the presence of R-VCF when compared with P-VCF (12.5% vs 0.6%; P = .007). All observed vena cava thromboses were associated with severe clinical symptoms and occurred in patients who received opposed biconical VCF designs.In our experience, both P-VCF and R-VCF can be placed safely. Among both permanent and retrievable devices, however, opposed biconical designs seem to be associated with an increased risk for vena cava thrombosis. Although causative factors remain unclear, filter design and resultant flow dynamics may play an important role, because all episodes of vena cava thrombosis occurred in patients with a single-filter design.