We have used second differential near infrared spectroscopy of water to determine the mean optical path length of the neonatal brain. By obtaining the ratio of the second differential features of deoxyhemoglobin to those of water, the absolute cerebral concentration of deoxyhemoglobin can be monitored continuously and noninvasively. Nineteen neonates were studied; the gestational age at birth varied from 23 to 38 wk, and the postconceptual age, when the spectra were recorded, ranged from 35 to 48 wk. The calculated mean deoxyhemoglobin concentration was 14.6 +/- 4.0 microM; the differential path length factor (mean optical path length/optode separation) calculated from the water peak at 730 nm was 4.66 +/- 1.01, and that calculated at the 830-nm peak was 3.91 +/- 0.75. These values are consistent with path length measurements using laser time-of-flight spectroscopy on postmortem neonates and phase-resolved spectroscopy on live neonates. Induced arterial oxygen saturation decreases from 98 to 93% showed no significant change in the mean optical path length, despite significant cerebral desaturation. Changes in the deoxyhemoglobin concentration after this procedure were identical, whether measured by second differential analysis at 760 nm or by multilinear regression over the wavelength range 740-900 nm. When combined with existing methods of measuring total cerebral hemoglobin concentration, second differential near infrared spectroscopy can be used to derive the mean cerebral oxygen saturation. A preliminary experiment outlined the feasibility of this approach and yielded a saturation value of 63%, consistent with near infrared sampling of a predominantly venous pool in the brain.