We have measured the ventilatory response to inhaled CO2 of six newborn infants in rapid-eye-movement (REM) and non-REM (NREM) sleep. Ventilatory responses were measured using the Read rebreathing technique. The response was further partitioned into the volume contributions of the rib cage and abdominal compartment using the respiratory inductance plethysmograph. Sleep state was defined by electroencephalogram, electrooculogram, and behavioral criteria. In NREM sleep, there was a highly significant linear correlation between both tidal volume (VT) and instantaneous minute ventilation (VI) with CO2. Among infants, the slope of VT varied from 1.0 to 0.34 ml X Torr-1 X kg-1. However, these differences were largely due to differences in rib cage contribution, which varied from 0.56 to -0.08 ml X Torr-1 X kg-1. The abdominal contribution was similar among infants (0.41-0.56 ml X Torr-1 X kg-1). In REM, the slopes of VI were less steep than in NREM, with greater breath-to-breath variability. Slopes of VT also tended to be lower. The abdominal responses were similar to those in NREM, whereas the rib cage response was low and negative in three studies. We conclude that the slope of the CO2 response curve is primarily determined by the extent of rib cage recruitment.