The success of noninvasive preoperative evaluation of infants with congenital heart disease using cardiac ultrasound depends not only on diagnostic accuracy, but also on risk of morbidity and mortality as compared with infants who undergo cardiac catheterization. Fifty-six infants (age 10 weeks or younger) with coarctation of the aorta (n = 16), coarctation with ventricular septal defect (n = 12), valvar aortic stenosis (n = 10) or total anomalous pulmonary venous connection (n = 18) were examined. Thirty-one underwent noninvasive preoperative assessment and 25 underwent evaluation including cardiac catheterization. Age, level and duration of support, pH, renal function, mortality, complications of cardiac catheterization and errors of diagnosis were compared. Significant differences between the 2 groups were more frequent preoperative use of prostaglandin E1 and shorter hospital stay in the noninvasively evaluated coarctation group. Of the infants with coarctation and ventricular septal defect, 1 who had cardiac catheterization required renal transplantation and 1 evaluated noninvasively required surgery at age 3 months for mitral stenosis not discovered on preoperative evaluation. One noninvasively evaluated infant with total anomalous pulmonary venous connection had a stenotic communication between the pulmonary venous confluence and the left atrium not detected by ultrasound. Surgery was successful in the latter 2 infants. Noninvasive preoperative diagnosis of some infants with congenital heart disease can be performed without increasing the risk of operative morbidity and mortality. Eliminating cardiac catheterization reduces hospital costs, decreases total numbers of catheterizations performed and influences the structure of training programs.