Exercise cardiac output (Q) was determined using the CO2 rebreathing equilibrium method. Five repeat tests in 12 boys and two tests over a 4 month interval in 47 boys were performed. Regression equations to predict Q from VO2 were in close agreement with dye dilution studies in boys (Eriksson and Koch 1973). Group mean data were reproducible from trial to trial. The day-to-day variability of Q, with a coefficient of variation of 7-8%, was found to be higher than when the CO2 method has been applied in adults. This greater variability was related, in part, to a larger biological variation in children as depicted in such relatively simple measures as submaximal exercise heart rate. The larger variability was also related to inaccuracies in the methods of PaCO2 estimation in children. Estimation from end-tidal CO2 concentrations requires further research to establish a correction for the alveolar-arterial gradient during exercise in children. Estimation of the child's dead space in exercise, with subsequent derivation of PaCO2 from the Bohr equation, also could be improved. Nevertheless, Q estimates in children exercising above VO2 1.01 X min-1 showed a day-to-day and long term stability acceptable for use in research and clinical studies.