We used single inspiratory capacity breaths of 5, 6 or 8% CO2 in air to obtain ventilatory responses in normal subjects, with ensemble averaging of repeated runs to define stimulus and response (Protocol 1). We also compared the effect of an inspiratory capacity (IC) breath of 8% CO2 with that of two tidal volumes (TV) at the same concentration (Protocol 2). The ventilatory response was defined first as the ratio of peak changes in ventilation and end-tidal PCO2, and secondly by the ratio of their integrals. We obtained group mean values of 0.21 L min-1 mmHg-1 for the peak method and 0.80 L min-1 mmHg-1 for integrals (Protocol 1). There was no significant difference between IC and TV response values (Protocol 2) either by the peak method (0.17 vs 0.19 L min-1 mmHg-1) or by integrals (0.47 vs 0.53 L min-1 mmHg-1). A significant decrease in ventilation was seen in the second tidal volume 8% CO2 breath, even though the stimulus was unperceived by four out of five subjects. CO2 responses can be obtained from these techniques, but the necessary analysis is too cumbersome for general use. Taking a deep breath had no detectable separate effect, but CO2 in the airway may depress ventilation even at concentrations which the subject cannot detect.