Ventilatory management patterns in very low birth weight newborns, particularly iatrogenic hypocapnia, have occasionally been implicated in perinatal brain damage. However, such relationships have not been explored in large representative populations. To examine the risk of disabling cerebral palsy in mechanically ventilated very low birth weight infants in relation to hypocapnia and other ventilation-related variables, we conducted a population-based prospective cohort study of 1105 newborns with birth weights of 500-2000 g born in New Jersey from mid-1984 through 1987, among whom 777 of 902 survivors (86%) had at least one neurodevelopmental assessment at age 2 y or older. Six hundred fifty-seven of 777 assessed survivors (85%), of whom 400 had been mechanically ventilated, had blood gases obtained during the neonatal period. Hypocapnia was defined as the highest quintile of cumulative exposure to arterial PCO(2) levels <35 mm Hg during the neonatal period. Disabling cerebral palsy was diagnosed in six of 257 unventilated newborns (2.3%), 30 of 320 ventilated newborns without hypocapnia (9.4%), and 22 of 80 ventilated newborns with hypocapnia (27.5%). Two additional ventilatory risk factors for disabling cerebral palsy were found-hyperoxia and prolonged duration of ventilation. In a multivariate analysis, each of the three ventilatory variables independently contributed a 2- to 3-fold increase in risk of disabling cerebral palsy. These risks were additive. Although duration of mechanical ventilation in very low birth weight newborns likely represents severity of illness, both hypocapnia and hyperoxia are largely controlled by ventilatory practice. Avoidance of arterial PCO(2) levels <35 mm Hg and arterial PO(2) levels >60 mm Hg in mechanically ventilated very low birth weight infants would seem prudent.