In adults, breath-by-breath analysis has been used for measuring respiratory gas exchange during exercise. The present study evaluates the validity and reproducibility of this method in children. In 21 patients with various types of congenital heart disease, steady state exercise testing was performed on a motor-driven treadmill. Based on simultaneous measurements of VO2, VCO2, VE and R, comparisons were made between the breath-by-breath and Douglas bag methods. No significant differences were found between both methods for any of the variables. In seven other patients the reproducibility of cardiorespiratory variables during exercise was assessed. No significant difference was found for the cardiorespiratory variables during any of the tests and the coefficients of variation were comparable to data obtained in adults. It is concluded that the breath-by-breath method for measuring respiratory values can be applied in children with an acceptable degree of validity and reproducibility.