Several devices are available for transcatheter occlusion of atrial septal defect. This report describes the international experience with the buttoned device. During a 4.5-year period ending in February 1993, 180 transcatheter atrial septal defect occlusions were performed with the buttoned device. Patient age varied between 0.6 and 76 years and stretched atrial defect diameter between 5 and 25 mm. The defects were closed with 25 to 50 mm devices delivered through 8F (148 patients) or 9F (32 patients) sheaths. Twelve patients were adults whose defects were closed to prevent recurrence of cerebrovascular accidents caused by presumed paradoxic embolism. In the remaining patients the atrial defect was closed to treat the left-to-right shunt. The atrial septal defects were effectively occluded as demonstrated by (1) decrease in pulmonary-to-systemic flow ratio from 2.1 +/- 0.6 (mean +/- SD) to 1.05 +/- 0.1 (p < 0.01) by oximetry; (2) normalized S2 and disappearance of the diastolic murmur by auscultation; and (3) improvement in right ventricular volume overloading by echocardiogram. However, trivial to small shunts could be detected by color Doppler studies in 76 (45%) of 168 patients in whom such data are available. Complications included unbuttoning in 13 and whole-device embolization in 1. All patients remained stable, and retrieval of the device and surgical closure of the atrial septal defect were accomplished in 10 patients. Transcatheter retrieval was used in the remaining 4 patients. The incidence of unbuttoning, a major complication of the procedure, appeared to decrease with the increasing experience of the investigators and with device modification (third-generation). The follow-up duration varied between 1 month and 4 years. Six patients required surgery during the follow-up period. In the remaining patients (n = 160), clinical examination did not reveal signs of atrial shunts. Color Doppler studies revealed either complete disappearance of the previously demonstrated shunts or further diminution of their size. The results indicate that transcatheter occlusion of the atrial septal defects with buttoned devices is feasible, relatively safe, and effective, and it appears to be a viable alternative to surgery for some patients with secundum atrial septal defect. Complications are infrequent and should improve with experience.