Near-infrared spectrophotometry can be used to measure cerebral concentrations of oxyhemoglobin and deoxyhemoglobin. This has been applied to developing methods for quantifying cerebral blood volume (CBV), which is relevant for the investigation of the pathogenesis of brain injury in newborn infants as well as older infants. This study investigates the internal consistency between measurements of CBV using two methods: the oxygen method, which is able to determine absolute values of CBV, and the total Hb method, which can detect changes in CBV only. Cerebral blood flow (CBF) was also measured. Fifteen premature infants were examined. Due to practical problems, in only eight of these was a minimum of two CBF and two CBV values obtained both before and after a change in arterial PCO2 of at least 0.5 kPa. A significant difference between the CBV-CO2 reactivity found by the two methods was demonstrated: 0.89 mL/100 g/kPa (95% confidence interval = 0.63-1.26) for the oxygen method and 0.22 mL/100 g/kPa (95% confidence interval = 0.08-0.36) for the total Hb method. This finding is substantiated by the absolute values of CBV [mean value = 3.7 mL/100 g (SD = 1.1)], CBF [mean value = 11.3 mL/100 g/min (SD = 5.9)], and CBF reactivity [59 +/- 9% (SEM)]. All the values correspond well with previous findings, although the CBV reactivity determined by the oxygen method has not been reported previously. The reason for the discrepancy between the two methods is unclear, but induced changes in the scattering properties of the brain would give rise to errors influencing the total Hb method rather than the oxygen method.