In contrast to shunt implantation, the indication for endoscopic ventriculostomy in patients diagnosed for normal-pressure hydrocephalus (NPH) is not scientifically established. Between September 1997 and May 2002, NPH was diagnosed in 91 patients. The diagnosis was established by means of the clinical signs, the intrathecal lumbar or ventricular infusion test, the cerebrospinal fluid tap test, and magnetic resonance imaging (MRI) cerebrospinal fluid (CSF) flow studies preoperatively and postoperatively. In 70 of these patients (77%), a ventriculoperitoneal shunt with a Miethke dual-switch valve (MDSV) was implanted, and in 17 patients (19%), endoscopic-assisted ventriculostomy was performed. Four patients refused an operation and were excluded from this study. With our own created NPH recovery rate and use of the clinical grading system for NPH created by Kiefer and his colleagues, the operative results in an average time interval of 12 and 27 months were compared in both groups of patients. In the group of patients with a shunt operation, 10 surgical revisions (14%) were necessary because of 4 shunt infections (6%), 2 shunt obstructions (3%), 2 cases of overdrainage (3%), and 2 catheter dislocations (3%). In the ventriculostomy group, the complications were pneumatocephalus in 1 patient (6%) and a partial ischemic thalamus lesion with temporary clinical signs in another patient. Underdrainage was seen after both surgical procedures: after shunt operation in 3 patients (4%) and after ventriculostomy in 3 patients (18%). In these patients, we changed the valve and inserted one with a lower opening pressure or implanted an MDSV in 2 patients after the ventriculostomy. In another patient, we performed an additional ventriculostomy. In patients with a pathologically increased resistance to CSF outflow in the lumbar infusion test caused by NPH, shunt implantation is indicated. Patients whose outflow resistance is increased in the ventricular infusion test but who are suspected of having functional aqueduct stenosis as the result of a physiologic lumbar infusion test should be treated by means of endoscopic ventriculostomy, and they should be monitored with clinical and neuroradiologic long-term follow-up. In case of inadequate experience with ventriculostomy or in cases with an uncertain result of the intrathecal infusion test, the shunt operation with implantation of a hydrostatic valve is the method of first choice.