A clinical trial was conducted of the buttoned device for transcatheter closure of ostium secundum atrial septal defect. Occlusion was attempted in 57 patients aged 1 to 62 years (median 5). The procedure was abandoned in 7 patients after 1 or more unsuccessful attempts, and devices were released in 50 patients. Urgent surgical retrieval was necessary in 4 patients because of unstable device position: 3 devices "unbuttoned," with migration of the counteroccluder to the pulmonary artery or inferior vena cava, and 1 intact device embolized to the main pulmonary artery. All patients remained stable and underwent successful operation. Successful device implantation was therefore achieved in 46 patients, with immediate reduction in Qp/Qs from 1.8 +/- 0.6 to 1.1 +/- 0.2 (mean +/- SD, p < 0.0001). At most recent follow-up (1 to 20 months), 45 of 46 patients (98%) have no shunt or a trivial residual shunt. The prevalence of residual shunts declined from 65% at 1 month to 19% at 12 months after the occlusion procedure (p < 0.0001). Complications included unbuttoning of a fourth device, transient tricuspid regurgitation in 2 patients, and transient mitral regurgitation in 2 patients. An episode of asymptomatic atrial flutter was noted in a 46-year-old patient which may have been related to device implantation, but which has not recurred. There have been no cases of endocarditis or thromboembolism in 350 patient-months of follow-up. The buttoned device provided effective closure of the atrial septal defect in 45 of 57 patients (79%) in whom implantation was attempted and in 45 of 46 patients (98%) in whom device implantation was successfully accomplished.(ABSTRACT TRUNCATED AT 250 WORDS)