Pulsatile flow in the anterior cerebral arteries was studied and pulsatility index (PI) calculated by 3 observers in 10 newborn infants in order to establish the reproducibility of a noninvasive Doppler technique. No significant differences were noted among the observers and the estimated error of observation was 0.073. Subsequently, 14 healthy preterm infants were studied daily by a single observer in an attempt to establish a normal range of PI for prematures in the first 5 days of life. The babies were concomitantly studied with serial ultrasound examinations to rule out intracranial hemorrhage. The study revealed good interobserver reliability but wide variation of PI in normal, healthy premature infants with values falling within abnormal ranges previously described in asphyxiated infants or those with intraventricular hemorrhage (IVH). Caution is advised in using the PI to predict outcome.