Regional cerebral blood flow (rcbf) was studied preoperatively and at 2 and 6 months postoperatively in 22 normal pressure hydrocephalus patients using xenon-133 inhalation and single photon emission computed tomography. Sixteen of the 22 patients improved (improved group) and six did not (unimproved group). The following comparisons were made: (1) preoperative rcbf in the improved group, to 14 normal elderly volunteers and to that in 59 SDAT (senile dementia of the Alzheimer type) patients; (2) preoperative rcbf in the improved and unimproved groups to determine if rcbf could predict surgical outcome; (3) pre- to postoperative rcbf in the improved group to see if increased cbf accounted for clinical improvement. The findings were: (1) preoperative rcbf in the improved group was lower than that in normal controls but was the same as that in SDAT; however, the ratios of rcbf values in anterior and posterior brain regions were significantly different between improved group and SDAT (p = 0.02); (2) an anterior/posterior ratio of 1.05 correctly classified surgical outcome in 19/22 patients; five of six in the unimproved group were above this cut off while 14/16 in the improved group were below; (3) in the improved group rcbf increased at 2 but not at 6 months after surgery without a corresponding reduction of clinical signs, supporting the notion that increase in cbf probably does not account for clinical improvement in normal pressure hydrocephalus.