Seventy-five paired comparisons of expired carbon dioxide tension (PECO2) and PaCO2 were obtained to evaluate the clinical usefulness of a mass spectrometer for monitoring PECO2 in critically ill neonates and to explore the factors that influence the PaCO2-PECO2 gradient. Nine neonates with a mean birth weight of 2.18 kg and a mean gestational age of 34 wk were studied during continuous positive-pressure ventilation. When PaCO2 was between 26 and 50 torr, this variable was linearly related to PECO2 by the equation: PaCO2 = 23.42 + .50 PECO2, r = +0.69 (p less than .001). Mean daily gradients for severe pulmonary disease (n = 9; mean 10.5 +/- 4.4 [SD] torr) were significantly (p less than .01) higher than those for moderate disease (n = 10; mean 3.9 +/- 3.1 torr); however, 61.3% of the gradients were no greater than 10 torr. Expiratory times no greater than 1.0 sec were associated with significantly (p less than .001) higher gradients (n = 33; mean 12.2 +/- 3.8 torr) than those over 1.0 sec (n = 42; mean 5.4 +/- 4.9 torr). This was, in part, due to the response time of the mass spectrometer system (0.68 sec to 63% of a step change in CO2). The continuous monitoring of PECO2 was especially helpful in the care of two neonates. We hypothesize that a decrease in the response time would improve the correlation between PaCO2 and PECO2 and thereby enhance the usefulness of this device as a trend monitor.