Cardiac output (Q) was estimated in supine rest and in upright cycling at several work rates up to 200 W in five male and one female subjects. At least four repetitions of both the CO2-rebreathing plateau method (Collier, J. Appl. Physiol. 9:25–29, 1956) and the Kim et al. (J. Appl. Physiol. 21: 1338–1344, 1966) single-breath method were performed at each work rate, in a steady state of O2 consumption and heart rate. At supine rest and low work rates, estimates of Q were similar by the two methods. However, at higher work rates, the single-breath method significantly (P less than 0.05) underestimated the value obtained by CO2 rebreathing. The reason for the difference in estimates of Q by the two methods was traced to the determination of arterial partial pressure of CO2 (PaCO2) and mixed venous partial pressure of CO2 (PvCO2). The estimate of PaCO2 from the single-breath method was approximately 88.5% of the estimate from end-tidal PCO2 used with the rebreathing method (P less than 0.001). The oxygenated PvCO2 calculated from the single-breath Q averaged approximately 92.5% of the PvCO2 from CO2 rebreathing (P less than 0.0001). The difference in estimates of Q was not eliminated by using a logarithmic form of the CO2 dissociation curve with the single-breath method.