We tested the hypothesis of whether the reduced ventilatory response to CO2 in preterm as compared to term infants is related to primary central unresponsiveness, or to mechanical impairment of the respiratory pump. Eleven preterm (n = 19; gestational age 32 +/- 0.4 wk) and 14 term (n = 24; GA 40 +/- 0.3 wk) infants were studied. Minute integrated diaphragmatic activity EMGDi X f), and mean inspiratory diaphragmatic activity (EMGDi/TI), were used as indices of central output. After 3 min breathing 21% O2 (control), infants rebreathed from a bag containing 5% CO2 in 40% O2 for 2 to 3 minutes. We measured VE, VT, f, VT/TI. Sleep states were monitored. Preterm infants had a decreased ventilatory response to CO2 both in quiet sleep (QS) (0.0379 +/- 0.067 vs 0.505 +/- 0.032 L . (min . kg . kPa PACO2)-1; P less than 0.04) and in active sleep (AS) (0.210 +/- 0.032 vs 0.331 +/- 0.048 L . (min . kg . kPa PACO2)-1; P less than 0.04). The decrease in response primarily was a function of a lack of increase in tidal volume with CO2 in QS and a lack of increase in f in AS. Parallel to these changes there were significant correlations between the increases in EMGDi X f and VE with inhaled CO2 (r = 0.75; P less than 0.001); VT and EMGDi (r = 0.63; P less than 0.01); and between the increases in EMGDi/TI and VT/TI with inhaled CO2 (r = 0.64; P less than 0.001). The results suggest that ventilatory response to CO2 is (1) correlated highly with diaphragmatic indices of central output; (2) less in active than in quiet sleep; (3) less in preterm than in term infants. We conclude that despite their increased chest wall compliance, preterm infant respond less to CO2 because of central unresponsiveness.