Experiments were conducted in ten adult men to determine if rapid-eye-movement sleep (REMS) reduced the ventilatory response to two 'steady-state' respiratory loads compared to slow wave sleep (SWS). A constant addition of 150 (or 200) ml/min pure CO2 to the inspirate (7 subjects) and 230 ml of added dead space (5 subjects) were the two respiratory loads. Inspiratory ventilation was measured by pneumotachygraph for at least five continuous minutes in wakefulness, SWS and REMS. The increase in ventilation to both stimuli was equal in SWS and REMS with no suggestion of an impaired response during the latter: increases in ventilation during CO2 loading being 49 and 51%, SWS and REMS, respectively, and during additional dead space they were 53 and 59%, SWS and REMS, respectively. Following the addition of extra dead space, end tidal PCO2 levels did not rise significantly more during REMS compared to SWS (P greater than 0.5). These findings show that when ventilatory responses to CO2 are considered across the whole of REMS (including periods with and without actual eye-movements) then they do not appear to be blunted compared to SWS.