Ventriculo-peritoneal shunt malfunction may be caused by shunt infection which may not be clinically apparent as the cause of the malfunction by standard diagnostic criteria. This suggests that the real incidence of infected shunts might be higher than previously suspected. In order to study the relationship between infection and shunt malfunction, we followed a protocol over five years (54 V-P shunts) consisting of (1) removal of the malfunctioning shunt and replacement in the same surgical procedure with a new one or institution of an external ventricular drainage for 8 days (if there were clear signs of infection), (2) culturing of CSF and every part of the removed shunt, and (3) intravenous antibiotic treatment (Vancomycin 1g./12h + Ceftriaxone 1g./12h) for five days after the new V-P shunt had been inserted. In those cases in which an external ventricular drainage had been placed, its tip and a portion of the new V-P shunt were also cultured. The results showed that although CSF cultures were negative in 49/54 cases (90.7%), cultures of the removed shunts were positive in 32/54 (59.2%), most of them (21/32, 65.6%) for Staphylococcus coagulase negative organisms. The CSF samples obtained by puncturing the reservoir on admission to Hospital were positive only in 5 out of 54 cases (9.2%), only in those showing clinical features of infection. In the remaining cases, 27 out of 54 (50%) the CSF cultures were negative but the shunt cultures proved positive and required further treatment. For the newly inserted shunts (173) CSF was collected through the shunt during the surgical procedure, and a small piece of the extra-tube from the ventricular and from the peritoneal catheter were obtained and cultured. All the six shunts (6/173, 3.4%) that showed positive cultures after insertion had to be replaced within a period of three to four weeks due to malfunction (range 26 +/- 7 days), indicating that the systematic culture of CSF and tubing helps to predict which shunts will soon need to be replaced due to infection. We conclude that CSF culture alone does not rule out infection in cases of shunt malfunction. The percutaneous CSF obtained from the shunt reservoir admission is particularly prone to show negative cultures even when the shunt is colonized by bacteria.