To evaluate the accuracy of noninvasive determination of stroke volume in infants and children, 28 patients (age range 4 weeks to 19 years) were studied. Stroke volume was calculated according to Teichholtz from M-mode echocardiographic tracings of left ventricular dimensions in 8 subjects. Agreement with thermodilution performed within 60 min of echocardiography was good (r = 0.995, y = 0.91x + 1.59, SEE = 1.8 ml). Since stroke volume correlated to body size we corrected for (height)3. After this correction there was still good agreement to thermodilution (r = 0.88, y = 1.29x-7.13, SEE = 7.1 ml/H3). M-mode echocardiography was then used as a reference method for evaluating two different Doppler methods in the remaining 20 subjects. Continuous wave Doppler stroke distance, calculated from the mean velocity, was combined with aortic root area (Method I), and stroke distance calculated from maximum velocity was combined with the aortic interleaflet area (Method II). Good agreement was found with Method I (r = 0.95, y = 1.01x-0.14, SEE = 8.1 ml) and Method II (r = 0.95, y = 1.04x-1.14, SEE = 8.4 ml). However, when stroke volume was normalized for (height)3, Method I was found to be superior to Method II.