Forty-two preterm infants of 28-33 weeks of gestation were studied once during the first week of life by 133-Xenon clearance after intravenous injection to estimate global cerebral blood flow. Count rates detected over the chest were corrected for chest wall contribution and used as arterial input function. A neonatal blood-brain partition coefficient of Xenon was used for the calculation of a mean flow estimator (CBF-infinity). The technique was internally validated by use of differently obtained arterial input functions. In 11 infants without respiratory distress, CBF-infinity was 19.8 ml/100 g/min +/- 5.3 SD. In 24 infants treated with mechanical ventilation CBF-infinity was 11.8 ml/100 g/min +/- 3.2 SD. In 7 infants treated with continuous positive airway pressure CBF-infinity was 21.3 ml/100 g/min +/- 12.0 SD. When the reduction of CBF-infinity associated with mechanical ventilation was taken into account, the 9 infants with subependymal/intraventricular haemorrhage had increased CBF-infinity. The effects of gestational age, birthweight, mode of delivery, postnatal age, mean arterial blood pressure, PaCO2, blood haemoglobin and phenobarbitone medication were also analysed and found inconsistent. In conclusion, CBF was lower than expected and in infants requiring mechanical ventilation the values were lower still.