We wished to test the hypothesis that end-tidal CO2 (ETCO2) is a function of tidal volume (Vt) in preterm infants requiring assisted ventilation (IMV). Optimal Vt would be associated with the highest ETCO2 and the lowest dead-space/tidal volume (Vd/Vt) ratio. We predicted an insufficient Vt would underventilate the patient and result in a low ETCO2, while an excessive Vt would overdistend the lungs, increase physiological dead space, and also lower ETCO2. Arterial blood was sampled to determine arterial PaCO2. ETCO2 was next sampled via a side-stream infra-red analyzer from the infant's endotracheal tube. A simultaneous recording was made of both spontaneous and mechanical breaths, and ETCO2 was correlated with Vt for a run of 30 consecutive breaths, usually over a 2-3 fold range of Vt. PaCO2 and ETCO2 were used to calculate arterial-alveolar CO2 differences and Vd/Vt ratios. 29 infants with an average birthweight of 1102 +/- 266 grams and a mean gestational age of 27.5 +/- 2.0 weeks were tested on 73 occasions between 2 and 27 days. The average alveolar-arterial O2 difference was 34.0 +/- 6.6 mm Hg; the average arterial-alveolar CO2 difference was 7.6 +/- 5.8 mm Hg. Mean Vd/Vt ratio for all studies was 0.19 +/- 0.13. A highly significant correlation between Vt and ETCO2 (r2 >.20) was found on 34/73 tests in 11/29 babies. This relation has the form of an inverted parabola, ETCO2 = 11.4 + 8.7(Vt) - 1.1(Vt)2. These results indicate that ETCO2 can be used to determine an optimal Vt associated with a minimal Vd/Vt ratio, avoiding overinflation or underventilation in a high percentage of infants requiring IMV.