1. The transient ventilatory response to CO2, measured using short pulses at constant inflow, was compared with the steady-state response at rest and during exercise at 50 W, and with the rebreathing response at rest, in nine healthy subjects. At rest CO2 was given at flow rates of 0.2 and 0.4 l min-1 and during exercise, to compensate for the smaller inhaled CO2 fraction as ventilation increased, at flow rates of 0.4 and 0.8 l min-1. 2. We calculated two indexes of gain for the transient response: the ratio of the peaks of the ventilation and PCO2 pulses, and the ratio of their integrals. 3. The steady-state response was greater than the transient response at rest and during exercise, but there was no correlation between the two. The rebreathing response was greater than both. Both the transient and the steady-state responses were greater during exercise than at rest. 4. To assess alinearity, the steady-state responses to the two CO2 flow rates were compared. At rest, there was no significant difference. During exercise, the response was greater to 0.4 than 0.8 l min-1, indicating alinearity concave downwards. 5. We conclude that the transient response as we calculate it is not representative of steady-state gain, and that the CO2 response in light exercise is steeper, and concave downwards in shape. The rebreathing technique overestimates CO2 sensitivity near the control point.