Ventricular shunts that require multiple revisions are familiar to pediatric neurosurgeons. We conducted a retrospective study to determine whether patients who require repeated shunt revisions represent a particular cohort within shunted hydrocephalic children. The clinical records of 244 children who had undergone shunt procedures between January 1990 and January 1996 were examined. They were divided into group 1: children with no shunt failure (n=136), group 2: children with one shunt revision (n=52), group 3: children with 2 or 3 shunt revisions (n=34), and group 4: patients who had 4 or more shunt revisions (n=22). Patients in groups 3 and 4 accounted for 54.8% of the total of 531 shunt procedures. Etiology of hydrocephalus, nature of the dysfunction, CSF characteristics, and variables related to the surgical procedure were analyzed for each group. We observed a progressive shortening of the intervals between revisions as the numbers of surgeries increased, indicating that shunts that tended to fail repeatedly did so sooner than those that did not. A Kaplan-Meier shunt survival curve showed that group 2 had a slower rate of failure than either group 3 (chi2=7.13, P<0.01) or group 4 (chi2=4.76, P<0.05). The etiologies of the hydrocephalus were not randomly distributed among the four groups (chi2=81.4, P<0.001); there was a predominance of congenital conditions in group 1. Repeated shunt revisions were associated with a progressive increase in the concentration of monocytes in the CSF (Kruskal-Wallis, P<0.05). Our data suggest that multiple shunt revisions constitute a phenomenon that may be caused by specific, still unidentified, biological factors.