Background: Use of Greenfield filters (GFs) to prevent fatal pulmonary embolism (PE) in trauma patients is generally well accepted. Nonetheless, a surprisingly small number of trauma surgeons insert filters in their patients. Among the reasons cited is fear of complications. Methods: We observed three femoral arteriovenous fistulae (AVF) in trauma patients who had inferior vena caval placement of filters for PE prophylaxis in one 12-month period (academic year 1999). In an effort to document the magnitude of this problem, we evaluated trauma patients who had a GF inserted in academic year 2000. Results: During that year, 133 consecutive patients (8.6% of trauma admissions) received 133 GFs through a percutaneous approach. The most common isolated indications for GF insertion included closed head injuries (n = 28), multiple long bone fractures (n = 27), pelvic and acetabular fractures (n 6), spinal cord injuries (n = 16), and vertebral fractures (n = 3). Five patients had documented deep venous thrombosis (DVT) diagnosed by duplex ultrasonography before GF placement, and 11 patients had other indications requiring a filter. There were 37 patients with more than one indication requiring filter placement. Most patients (57%) underwent preinsertion duplex scanning of their lower extremity veins; 77% of patients underwent postinsertion scanning. Filters were inserted an average of 6.8 +/- 0.6 (SE) days after trauma. No AVF were suspected clinically or detected ultrasonographically. No operative or postoperative complications occurred. DVT was observed in 30% of patients despite 92% prophylaxis; there was a 26% incidence of de novo thrombi detected. None of the patients evidenced DVT clinically. Conclusion: Our data indicate that complications of GF insertion for prophylaxis against PE from DVT complicating trauma patients continue to be negligible. In addition, the incidence of insertion-site thrombosis may be lower than expected. Moreover, femoral AVF is a rare complication of this procedure.