1. The ventilatory sensitivity to carbon dioxide obtained from a step-ramp CO2 challenge was compared to the CO2 sensitivity from the steady-state method. 2. Experiments were performed in nine healthy male subjects against a background of hyperoxia and in two subjects against a background of normoxia. 3. In each subject experiments were performed in which the stepwise increase in end-tidal PCO2 above its resting value (A) was varied (range 0-2 kPa) and the subsequent rate of rise of end-tidal PCO2 in time (R) kept constant at 0.6 or 0.8 kPa min-1. 4. The results of the hyperoxic experiments show that the slope of the non-steady-state ventilatory response to CO2 (Sn) is greatly influenced by the magnitude of A. An increase of A of 1 kPa results in a 54% increase of the ratio non-steady-state ventilatory CO2 sensitivity to steady-state ventilatory CO2 sensitivity (Ss). The magnitude of R plays a minor role in determining Sn. The normoxic experiments gave similar results. 5. In experiments performed during hyperoxia Sn approximates Ss when the magnitude of A is 0.5 kPa. 6. The results are discussed and related to a physiological model. Simulations with representative values for the model parameters are in fair agreement with experimental values.