Percutaneous placement of caval filters is performed with increasing frequency as an alternative or adjunctive therapy for venous thromboembolic disease. The stainless steel Greenfield filter may be placed surgically or percutaneously. There are now several other approved or investigational devices primarily intended for percutaneous delivery. Morbidity and mortality rates for percutaneous filter placement are lower than those published for surgical filter placement. Choosing among the various devices on the basis of published studies is difficult, as the indications for filter placement; objectivity of follow-up; and criteria for complications, success, and failure vary greatly among studies. The author analyzes published and unpublished data about the physical characteristics; clinical availability; and morbidity, mortality, and efficacy rates of several caval filters. The author concludes that filters should be placed percutaneously rather than surgically, that the ideal device cannot be determined on the basis of current data (and might not exist), and that the choice among devices might depend on clinical circumstances and device characteristics. A strong recommendation is made for an independent, multicenter, prospective, randomized trial and, in the interim, a mandatory registry of all patients receiving caval filters.