We analyzed 3705 measurements of inspired gas temperature in the first 96 hours of life, with concurrent measurements of ventilator variables and arterial oxygen tension, to determine any relationship to respiratory sequelae in 149 infants assisted by artificial ventilation. All management was with one type of ventilator (model IV100B, Sechrist Industries Inc., Anaheim, Calif.) and one type of humidifier (model MR 500, Fisher and Paykel, Auckland, New Zealand), and infants were placed under radiant heaters. Outcome for infants weighing greater than or equal to 1500 gm at birth was no different at low (less than or equal to 36.5 degrees C) versus high (greater than 36.5 degrees C) values of mean inspired gas temperature during the first 96 hours. However, infants weighing less than 1500 gm at birth had less respiratory morbidity at the higher temperatures; there was a reduction in the incidence of pneumothorax from 43% at low to 13% at high temperatures (p = 0.006) and a reduction in the severity of chronic lung disease, measured as mean Fio2 at 29 days in survivors, from 37.2% at low to 27.5% at high temperatures (p = 0.001). Clinical evaluation of the humidifier, as it was used in this study, suggested that temperature settings less than 36.5 degrees C were associated with inspired gas humidity between 28 and 36 mg H2O/L. For any humidifier, there may be a critical threshold for inspired gas humidity below which the risk of respiratory complications in very low birth weight babies is increased. This hypothesis requires rigorous scrutiny by controlled trial. In future studies, direct measurements of inspired gas humidity are needed to make a precise estimate of the optimal level.