OBJECTIVE This prospective, randomized, controlled study was designed to investigate the safety, feasibility, and preliminary efficacy of long-term CSF drainage via a low-flow ventriculoperitoneal shunt in subjects suffering from AD. METHODS Twenty-nine subjects selected for probable AD (National Institute of Neurological and Communicative Diseases and Stroke-Alzheimer's Disease and Related Dementias Association criteria) were screened to exclude normal pressure hydrocephalus or other etiologies of dementia and randomized to treatment (shunt) or no treatment groups. The study endpoint was the comparison of group performance on psychometric testing at quarterly intervals for 1 year. Shunted subjects had CSF withdrawn for MAP-tau and Abeta((1-42)) assays at the same time intervals. RESULTS There was no mortality from the surgical procedure, and no patient sustained a subdural hematoma. Five notable postoperative adverse events, which resolved without permanent neurologic deficit, were reported in the shunt group. Group mean Mattis Dementia Rating Scale total scores showed little change over the year in the shunt-treatment group, in contrast to a decline in the control group (p = 0.06). Mini-Mental State Examination mean scores supported a trend in favor of shunt treatment (p = 0.1). There was a concomitant decrease in ventricular CSF concentrations of AD biomarkers MAP-tau and Abeta((1-42)). CONCLUSIONS The surgical procedure and the device are reasonably safe. Adverse events were consistent with shunt procedures for hydrocephalus in this older population. The endpoint data show a trend in favor of the treated group. A larger, randomized, double-blinded, controlled, clinical trial is underway.