The validity of a CO2 rebreathing (exponential) method to determine cardiac output was acceptable during exercise, but not at rest. The reproducibility during exercise of the measurement of cardiac output was similar to data published for the dye dilution method.