The pressure-volume index (PVI) technique of bolus manipulation of cerebrospinal fluid (CSF) was used to measure neural axis volume buffering capacity and resistance to absorption of CSF (Ro) in 20 shunt-dependent hydrocephalic children acutely ill from shunt malfunction. All children had had ventricles that were near normal or subnormal in size when the shunts were functioning. The mean intracranial pressure (ICP, +/- standard deviation (SD] at the time of revision was 10.6 +/- 6.4 mm Hg. The mean measured PVI (+/- standard error of the mean) was 18.4 +/- 1.1 ml compared to the normal PVI of 17.5 +/- 4.4 ml (+/- SD) predicted for these children. According to paired t-tests, these measured values were similar to those predicted on the basis of neural axis volume for each child, indicating that these children had normal neural axis volume buffering capacity. While the study was in progress, abrupt increases of ICP were documented in all children. These waves were observed spontaneously as well as in response to the addition of volume to the neural axis. In each child a specific threshold pressure along the pressure-volume curve corresponded to the appearance of unstable ICP. The threshold pressures at which this occurred corresponded to a mean neural axis compliance of 0.32 +/- 0.07 ml/mm Hg (+/- SD). The Ro varied as a function of ICP. The Ro measured at ICP's below 15 mm Hg ranged from 2 to 7.5 mm Hg/ml/min and rose to 12 to 30 mm Hg/ml/min at pressures in the 20 to 25 mm Hg range. The results of this study indicate that neural axis volume buffering capacity is normal in shunt-dependent children who respond to shunting by reconstitution of the cortical mantle. This study indicates that the proximate cause of their abrupt clinical deterioration is unstable ICP, which occurred at a similar point on the pressure-volume curve of all children studied. The correlation of Ro to ICP suggests that CSF absorption does not increase in these children as ICP rises, resulting in movement along relatively normal pressure-volume curves. The functional implications of these parameters are discussed.