This study is a retrospective analysis of intracranial pressure (ICP) and cerebral perfusion pressure (CPP) data from 56 children with active hydrocephalus and cerebrospinal fluid (CSF) shunt malfunction. The pressures were measured from a separately sited CSF access device placed in the frontal horn of the lateral ventricle. Of the patients, 79% had an elevated ICP (mean 20+/-12 mmHg). A subgroup of patients demonstrated ten different forms of CSF-filled swelling. This group had significantly lower ICP recordings (P=0.000075) with a mean ICP of 8.5 mmHg compared with the remainder (22.9 mmHg). This pressure 'compensation' was because of additional nonphysiological accommodation of CSF volume. Overall the CPP was normal in 35% of cases despite normal ICP occurring in only 11% of cases. The CPPs were not significantly different in those with and without compensation. Measurement of ICP may not always be a reliable indicator of shunt malfunction in shunt-dependent children who present with compensatory CSF-filled spaces.