We conducted this retrospective study to compare methods for measuring atrial septal defects and to identify factors affecting echocardiographic measurement of such defects before transcatheter closure with the CardioSEAL'Septal Occluder. We reviewed the records of patients considered for device placement at our institution from January 1997 to April 1999. Atrial septal defect size was measured by transthoracic and transesophageal echocardiography; the stretched diameter was measured during catheterization by fluoroscopy and transesophageal echocardiography. The stretched-diameter fluoroscopic measurement was used for device size selection. Analysis of variance was used to calculate the effect of size, age, and size-by-age interaction. Thirty-one patients (3.3 to 72 years of age) underwent transthoracic and transesophageal echocardiography One patient was excluded from catheterization because of a 25-mm septal defect as indicated by transesophageal echocardiography (our maximum diameter, 15 mm). Thirty patients underwent transcatheter stretched-diameter sizing; 5 were excluded from device implantation because of defects >20 mm by stretched-diameter fluoroscopy (4) or septal length insufficient for device support (1). Implantation was successful in 23/25 patients; 2/23 had a residual shunt. In patients with available results (26/30), the stretched diameter was the same whether measured by stretched-diameter fluoroscopy or transesophageal echocardiography (P=0.007 R square=0.963). Compared with stretched-diameter fluoroscopy, precatheterization transthoracic and transesophageal echocardiography underestimated defect size by a mean of 22% and 13.2%, respectively. When data from those same tests were compared in defects of < or =0 mm and > 10 mm, transthoracic and transesophageal echocardiography were reliable predictors (P=0.003 and P=0.05, respectively) of stretched-diameter size in defects < or =0 mm.