The accuracy of shunt calculations derived from Doppler echocardiography was assessed in 15 children (5-15 years, mean 8.5 yrs.) with secundum-type atrial septal defect and left-to-right (L-R) shunt. Qp/Qs was calculated by measuring stroke volumes over the aorta and pulmonary artery prior to and 4-15 days after corrective surgery. The measuring error of the Doppler method was defined as deviation of the postoperative shunt value from zero. A residual shunt was excluded by careful intraoperative testing and physical examination. Intraoperative inspection revealed normal pulmonary venous drainage. L-R-shunt, as calculated by Doppler echo, significantly decreased after surgery (p less than 0.001). Although all septal defects were closed, calculations by Doppler echo resulted in a median residual shunt of 16% (0-50.7%). IN 2/15 children a L-R shunt of more than 40% was derived by echo recordings. The size of the measuring error increased in children with great preoperative shunts. There was a significant correlation between the ratio of pulmonary to aortic cross-sectional area and the measuring error (p less than 0.03, r = 0.59). Poor estimation of cross-sectional areas in dilated pulmonary arteries was the most important limiting factor in shunt calculation. Doppler echocardiography provides a valuable noninvasive method for shunt calculation. Similarly to Fick's principle, in some patients the measuring error may be considerable and become relevant for clinical decisions. Thus Doppler-derived shunt calculations must be complemented by additional information of two-dimensional echo and physical examination.