The authors examined whether measurement of resistance to outflow of cerebrospinal fluid (Rcsf) predicts outcome after shunting for patients with normal-pressure hydrocephalus (NPH). In four centers 101 patients (most of whom had idiopathic NPH) who fulfilled strict entry criteria underwent shunt placement irrespective of their level of Rcsf obtained by lumbar constant flow infusion. Gait disturbance and dementia were quantified by using an NPH scale and the patient's level of disability was assessed by using the modified Rankin scale (mRS). In addition the Modified Mini-Mental State Examination was performed. Patients were assessed prior to and 1, 3, 6, 9, and 12 months after surgery. Primary outcome measures were based on differences between the preoperative and last NPH scale scores and mRS grades. Improvement was defined as a change measuring at least 15% in the NPH scale score and at least one mRS grade. Intention-to-treat analysis of all patients at 1 year yielded improvement for 57% in NPH scale score and 59% in mRS grade. Efficacy analysis, excluding serious events and deaths that were unrelated to NPH, was performed for 95 patients. Improvement rose to 76% in NPH scale score and 69% in mRS grade. Six cut-off levels of Rcsf were related to improvement in NPH scale score using two-by-two tables. Positive predictive values were approximately 80% for an Rcsf of 10, 12, or 15 mm Hg/ml/minute, 92% for an Rcsf of 18 mm Hg/ml/minute, and 100% for an Rcsf of 24 mm Hg/ml/minute. Negative predictive values were low. More important was the highest likelihood ratio of 3.5 for an Rcsf of 18 mm Hg/ml/minute. Extensive comorbidity was a major prognostic factor. Measurement of Rcsf reliably predicts outcome if the limit for shunting is raised to 18 mm Hg/ml/minute. At lower Rcsf values the decision depends mainly on the extent to which clinical and computerized tomography findings are typical of NPH.