BACKGROUND: The optimal strategy to prevent recurrent presumed paradoxical emboli in patients with patent foramen ovale is unknown. PURPOSE: To synthesize the current knowledge about and qualitatively assess the relative benefits of transcatheter closure versus medical therapy for patent foramen ovale. DATA SOURCES: English-language and foreign-language journals listed in the MEDLINE database from January 1985 to July 2003 were systematically searched. Secondary sources were also used. STUDY SELECTION: Secondary prevention studies of transcatheter closure or medical therapy for patent foramen ovale were required to include at least 10 patients followed for more than 1 year and to report recurrent neurologic events. DATA EXTRACTION: Data from published studies were manually extracted and summarized. DATA SYNTHESIS: Ten studies of transcatheter closure (1355 patients) and 6 studies of medical therapy (895 patients) for patent foramen ovale were included. Overall, the 1-year rate of recurrent neurologic thromboembolism with transcatheter intervention was 0% to 4.9%, and the incidence of major and minor complications was 1.5% and 7.9%, respectively. Medical management was associated with a 1-year recurrence rate of 3.8% to 12.0%. However, limitations resulting from uncontrolled data, nonstandardized definitions, vigilance of follow-up, and baseline imbalances preclude definitive conclusions about the superiority of a particular approach. General differences in study samples included older age, greater proportion of men, and higher prevalence of diabetes and smoking among medically treated patients. Patients undergoing treatment with a transcatheter device were more likely to have had multiple thromboembolic events at baseline. CONCLUSION: Transcatheter closure of patent foramen ovale may prevent a substantial proportion of cryptogenic strokes. Randomized clinical trials are needed.