Two estimations of global cerebral blood flow (CBF) using 133-Xenon clearance were done with an interval of about one hour in 16 mechanically ventilated, newborn infants, of less than 33 weeks gestational age. In eight infants CBF was estimated just before a change in ventilator settings, and again when the PaCO2 was stable. In the remaining eight infants small spontaneous changes in PaCO2 occurred. The CBF-CO2 reactivity was similar in the two groups (+67%/kPa (95% confidence interval 13-146) and 52%/kPa (24-86)) and considerably higher than the CBF-CO2 reactivity estimated from the interindividual variation of flow and PaCO2 (+19%/kPa (4-36)). There were no significant relations between CBF and arterial blood pressure. Flash evoked potentials (VEP) were recorded during the 133-Xenon clearances in 8 of the infants. VEP showed no relation to changes in CBF, even when the blood flow rose from the lowest levels. CBF and VEP were obtained once in 9 other infants. Among the 17 infants, the latency of the first negative wave of the VEP was not related to the CBF level. Mean CBF in the 25 infants was 12.3 ml/100 g/min (range 4.3 to 18.9), mean PaCO2 was 4.2 kPa (range 2.3 to 6.4). Thus, CBF-CO2 reactivity appeared to be normal in these clinically stable, mechanically ventilated, preterm infants, suggesting that their low cerebral blood flow was well regulated. The absence of a relation of CBF with VEP suggested that cerebral blood flow was not critically decreased.