There are two suitable methods for estimating the ventilatory response to CO2: the steady state or the Read method. The latter is usually applied because of its shorter duration and its stimulus, near to tissue PCO2. From recent studies on a physiological model but also on the dynamics of the ventilatory CO2 response (G liters min-1 kPa-1), it can be shown, both theoretically and experimentally, that the Read method markedly overestimates the steady-state estimate of the ventilatory CO2 response (Gss). We have, therefore, applied two ramp approaches: one with an initial end-tidal PCO2 (PetCO2) step of 0.5 kPa and a mean PetCO2 ramp slope of 0.41 kPa/min (step-ramp method, SR) and one without an initial step and a mean ramp slope of 0.21 kPa/min (ramp method; R). As predicted by theory, the ventilatory CO2 response from the R method (GR) during the second 4-min interval of the 8-min procedure and the ventilatory CO2 response from the SR method (GSR) from the whole 6-min procedure, except for the first 30 s, should yield a good approximation of Gss. In a group of 12 normal volunteers we indeed found no differences between these estimates, and, as predicted by theory, we found both estimates to be larger than GR from the first 4-min interval. From the interindividual spread, only the difference with respect to GSR was significant (p less than 0.02). The volunteers preferred the R approach because of the lower PetCO2 attained and the smaller ventilation increase.(ABSTRACT TRUNCATED AT 250 WORDS)