We evaluated 86 patients for possible normal pressure hydrocephalus (NPH) by: 1) CSF pressure (Pcsf) monitoring and analysis for percent of time with A or B-waves, and 2) controlled CSF drainage for 3 days via a lumbar subarachnoid catheter. Clinical outcome after CSF drainage and shunt surgery was assessed as change of clinical exam, with grades of none, minor, moderate, or marked change. For outcome analysis in 47 patients after shunt surgery, NPH was defined as moderate or marked clinical improvement. We assessed the diagnostic discrimination of percent-of-time thresholds of A and B-waves for 38 patients. At 10%, sensitivity for NPH is 91%, specificity is 13%, positive predictive value (PPV) is 62%, and the false positive rate is 38%. At the 25% threshold, sensitivity is 78%, specificity is 40%, PPV is 67%, false positive rate is 33%, and the false negative rate is 22%. For CSF drainage (threshold of minor improvement or better), the sensitivity is 97%, specificity is 60%, PPV is 84%, negative predictive value (NPV) is 90%, and the false negative rate is 3%. We conclude: 1) clinical response to controlled CSF drainage accurately predicts the outcome after shunt surgery in patients suspected of having NPH, and 2) A or B-waves poorly predict which patients will respond to shunt surgery. Three days of CSF drainage seems to encompass critical thresholds of CSF volume removal or duration of Pcsf reduction necessary for neuronal function to begin returning and symptoms to begin resolving in patients with NPH.