Between January 1976 and July 1983, 217 patients with atrial septal defect underwent surgical repair at Children's Hospital. Thirty with a primum atrial septal defect and 26 who underwent cardiac catheterization elsewhere before being seen were excluded from analysis. Of the 161 remaining patients, 52 (31%) underwent preoperative cardiac catheterization, 38 because the physical examination was considered atypical for a secundum atrial septal defect and 14 because of a preexisting routine indication. One hundred nine (69%) underwent surgery without catheterization, with the attending cardiologist relying on clinical examination alone in 5, additional technetium radionuclide angiocardiography in 5, M-mode echocardiography in 13 and two-dimensional echocardiography in 43; both M-mode echocardiography and radionuclide angiography were performed in 24 and two-dimensional echocardiography and radionuclide angiography in 19. Since 1976, there has been a trend toward a reduction in the use of catheterization and use of one rather than two noninvasive or semiinvasive techniques for the detection of atrial defects. Of the 52 patients who underwent catheterization, the correct anatomic diagnosis was made before catheterization in 47 (90%). Two patients with a sinus venosus defect and one each with a sinus venosus defect plus partial anomalous pulmonary venous connection, partial anomalous pulmonary venous connection without an atrial septal defect and a sinoseptal defect were missed. Of 109 patients without catheterization, a correct morphologic diagnosis was made before surgery in 92 (84%). Nine patients with a sinus venosus defect, three with sinus venous defect and partial anomolous pulmonary venous connection, four with partial anomalous pulmonary venous return without an atrial septal defect and one with a secundum defect were incorrectly diagnosed.(ABSTRACT TRUNCATED AT 250 WORDS)