Most removed shunt valves are discarded with no investigation into why they had to be replaced or whether they still performed to specification. However, improvements in valve design will only occur if valves that have needed removal are examined and the reasons that they failed are determined. An in-depth study of 43 valves that were removed in this unit over a 15-month period was performed. They were submitted to a four-part study, comprising flow-pressure testing, opening and closing pressure measurement, assessment of the susceptibility to syphoning, and dismantling with internal inspection of the valve components. Overall, 81% of valves failed to meet the manufacturers' specified performance data, even though the peroperative cause of shunt failure was thought to lie outside the valve. Over 80% of valves with metallic parts were found to have accumulated debris internally, and this was thought to have impaired their performance. In contrast, only 25% of non-metallic valves contained debris, a significant difference (0.01 > p > 0.001). All of the valves had a high tendency to overdrainage. Attention is drawn to the high number of malfunctioning valves. It is recommended that a change of valve should be considered in all shunt revisions, and that future valve designs should avoid metal components.