BACKGROUND Although it has been possible to reduce the percentage of premature infants suffering intraventricular hemorrhage, posthemorrhagic hydrocephalus remains a serious problem without a good treatment. There is a high rate of cerebral palsy, and ventriculoperitoneal shunt surgery makes the child permanently dependent on the valve and catheter system. Shunt surgery cannot be carried out early because of the blood in the cerebrospinal fluid (CSF) and the brain may be subjected to periods of raised pressure. Early tapping of CSF by lumbar puncture or ventricular tap was suggested as a way of temporarily reducing pressure and removing blood and protein and thereby avoiding permanent hydrocephalus. OBJECTIVE To determine whether repeated CSF tapping, by lumbar puncture or ventricular tap, reduced the risk of permanent shunt dependence, neurodevelopmental disability or death in neonates at risk of, or actually developing, post-hemorrhagic hydrocephalus (PHH). This form of treatment was based on the hypothesis that repeated tapping removed protein and blood from the CSF, thus clearing obstruction from the channels of CSF absorption. METHODS Pediatric, Neurosurgical and General Medical Journals were handsearched from 1976 up to October 2000, as well as the Medline database (via PubMed) and the Cochrane Controlled Trials Register. Personal contacts were used. METHODS Four controlled trials ( with five published papers) were identified, three being randomised and the fourth using alternative allocation. Two trials evaluated repeated lumbar punctures in neonates with intraventricular hemorrhage (IVH) and two trials evaluated repeated CSF tapping infants with IVH followed by progressive ventricular dilatation. METHODS In addition to details of the patient selection and patient allocation, the interventions were extracted. The end-points examined were: ventriculoperitoneal shunt, death, disability, multiple disability and death or disability. RESULTS The studies were sufficiently similar in the question they were asking and the interventions were sufficiently in common that they could be combined when assessing the effect of the intervention. When repeated CSF tapping was compared to conservative treatment, the relative risks for shunt placement, death, disability and multiple disability were very close to 1.0 with no statistically significant effect. There is also evidence that this form of treatment increased the risk of CSF infection. CONCLUSIONS Early repeated CSF tapping cannot be recommended for neonates at risk of, or actually developing, post-hemorrhagic hydrocephalus.