Recent experience with the spring-loaded patent ductus arteriosus (PDA) occluder has demonstrated several advantages of this device in the closure of intracardiac defects compared with previously described techniques. Pathologic and animal model studies were performed to identify which atrial septal defects (ASDs) might be suitable for device closure and to test a new septal closure double umbrella. Fifty specimens from the Cardiac Registry with unrepaired ASD secunda (2 degrees) were analyzed. Mean ASD size was 8 x 10 mm (range, 3 x 4 to 30 x 30 mm); 80% (n = 40) of ASDs were judged closable with the new device. ASD closure was attempted in four lambs with the Rashkind (hooked single umbrella) ASD occluder. One of four umbrellas was correctly positioned; no ASDs were closed. A new double-hinged ("clamshell") umbrella device was designed: eight ASD closures were attempted with this device (defects ranged from 8 to 20 mm in diameter). Six of eight umbrellas were correctly positioned; four of four animals observed more than 1 day appeared to have complete closure on postmortem examination with endothelialization of the device. We conclude that 1) most ASD 2 degrees are far enough from vital structures to permit closure, 2) initial placement of hooked umbrellas is often incorrect and cannot be altered, 3) clamshell double umbrellas were successfully placed in six of eight attempts, and 4) endothelialization of closed ASDs appears complete within weeks of implantation. These preliminary studies appear promising and support testing the clamshell ASD device in clinical trials.