Forty percent of standard cerebrospinal fluid shunts implanted for the treatment of pediatric hydrocephalus fail within the first year. Two new shunt valves designed to limit excess flow, particularly in upright positions, were studied to compare treatment failure rates with those for standard differential-pressure valves. Three hundred-forty-four hydrocephalic children (age, birth to 18 yr) undergoing their first cerebrospinal fluid shunt insertion were randomized at 12 North American or European pediatric neurosurgical centers. Patients received one of three valves, i.e., a standard differential-pressure valve; a Delta valve (Medtronic PS Medical, Goleta, CA), which contains a siphon-control component designed to reduce siphoning in upright positions; or an Orbis-Sigma valve (Cordis, Miami, FL), with a variable-resistance, flow-limiting component. Patients were monitored for a minimum of 1 year. Endpoints were defined as shunt failure resulting from shunt obstruction, overdrainage, loculations of the cerebral ventricles, or infection. Outcome events were assessed by blinded independent case review. One hundred-fifty patients reached an endpoint; shunt obstruction occurred in 108 (31.4%), overdrainage in 12 (3.5%), loculated ventricles in 2 (0.6%), and infection in 28 (8.1%). Sixty-one percent were shunt failure-free at 1 year and 47% at 2 years, with a median shunt failure-free duration of 656 days. There was no difference in shunt failure-free duration among the three valves (P = 0.24). Cerebrospinal fluid shunt failure, predominantly from shunt obstruction and infection, remains a persistent problem in pediatric hydrocephalus. Two new valve designs did not significantly affect shunt failure rates.