A clinical trial was conducted to assess the feasibility, safety, and efficacy of the atrial septal defect (ASD) occlusion system for transcatheter closure of secundum ASD and patent foramen ovale (PFO) after episodes of cerebral embolism. Occlusion was attempted in 200 patients aged 1 to 74 years (mean 32). The procedure failed in 26 patients (13%); the device was retrieved through a catheter in 20 and through surgery in 6 patients. Procedure-related complications necessitating surgical removal of the device included device embolization in 2, device entrapment within the Chiari network in 1, frame fracture in 1, and perforation of atrial wall in 2. All 6 patients experienced an uneventful postoperative course. An additional 11 patients (6%) underwent surgical removal of the device during follow-up. There were 163 patients (81%) with an implanted ASD occlusion system at follow-up of from 6 to 36 months (mean 17). Thrombus formation around the device was detected by transesophageal echocardiography in 9 patients 1 to 4 weeks after implantation. One of these patients (who had a coagulation factor XII deficiency) suffered a cerebral thromboembolism. Late atrial wall perforation (5, 6, and 8 months after implantation) occurred in 3 adult patients. Infectious endocarditis developed in 2 adult patients (1%). No late device embolization and no atrioventricular valve injury occurred. An asymptomatic device frame fracture was found in 14% and frame deformity in 4% of all patients during the follow-up period of >230 patient-years. Immediately after closure, a moderate/large residual shunt remained in 8% and a small shunt in 29% of patients. After 1 year, a moderate/large shunt was present in 2% and a small one in 26% of patients. During a total follow-up of 49 patient-years, only 1 of 46 patients with PFO had a transient neurologic event after the closure. The study indicates that patients with centrally situated secundum ASD and those with PFO after cerebral embolism can be treated with this system with a high success rate and an acceptable morbidity.