Twenty-seven patients with normal pressure hydrocephalus were operated upon by a ventriculo-peritoneal shunt. Selection for shunt surgery was based on typical symptoms (gait disturbancy, mental deterioration and urgency incontinence) and characteristic changes at cranial computed tomography and/or radionuclide cisternography. Prior to operation a cerebrospinal fluid tap-test (CSF-TT) was performed with measurements of psychometric functions and gait pattern before and after a lumbar puncture of 50 cc CSF. Nineteen patients improved and 5 were unchanged after shunt operation. Three patients could not be evaluated. Improvement in the psychometric functions and gait pattern after lumbar puncture correlated to improvement after the shunt operation (r = 0.64, p less than 0.01: r = 0.96, p less than 0.001, respectively). Improvement in 2 or more of the 4 tests used (3 psychometric and 1 gait test) at CSF-TT implied in all cases successful result of the shunt operation. It was concluded that CSF-TT could predict which NPH patient will improve by a shunt operation, and albeit to envisage the degree of improvement.