Aims: To evaluate the predictive value of symptoms, signs, and radiographic findings accompanying presumed ventriculoperitoneal (VP) shunt malfunction, by comparing presentation with operative findings and subsequent clinical course. Methods: Prospective study of all 53 patient referrals to a paediatric neurosurgical centre between April and November 1999 with a diagnosis of presumed shunt malfunction. Referral pattern, presenting symptoms and signs, results of computed tomography (CT) scanning, operative findings, and clinical outcome were recorded. Two patient groups were defined, one with proven shunt block, the other with presumed normal shunt function. Symptomatology, CT scan findings, and the subsequent clinical course for each group were then compared. Results: Common presenting features were headache, drowsiness, and vomiting. CT scans were performed in all patients. Thirty seven had operatively proven shunt malfunction, of whom 34 had shunt block and three shunt infection; 84% with shunt block had increased ventricle size when compared with previous imaging. For the two patient groups (with and without shunt block), odds ratios with 95% confidence intervals on their presenting symptoms were headache 1.5 (0.27 to 10.9), vomiting 0.9 (0.25 to 3.65), drowsiness 10 (0.69 to 10.7), and fever 0.19 (0.03 to 6.95). Every patient with ventricular enlargement greater than their known baseline had a proven blocked shunt. Conclusions: Drowsiness is by far the best clinical predictor of VP shunt block. Headache and vomiting were less predictive of acute shunt block in this study. Wherever possible CT scan findings should be interpreted in the context of previous imaging. We would caution that not all cases of proven shunt blockage present with an increase in ventricle size. C erebrospinal fluid (CSF) shunts remain the mainstay of treatment for most cases of hydrocephalus in the paediatric population. All are prone to malfunction, with block being the commonest reported complication in most series. In the largest reported cohort of 1719 patients, 56% experienced at least one episode of shunt block in the 12 years following insertion.1 Similarly, Lazareff and colleagues 2 recently reported a 44% prevalence of shunt malfunction, including block, in 244 children with CSF shunts followed up over a period of up to six years post initial insertion. The peak "danger" period for blockage is in the first year after insertion, with rates as high as 20% recorded in some series.3 Annual rates of shunt blockage have been estimated by Rekate to be approximately 5%. 4 These findings are in keeping with our published unit experience in which we noted a 28% incidence of shunt block over a 10 year period, and that 55% of patients experienced at least one episode of shunt malfunction during this time.