The radioisotope distribution following intravenous injection of 99mTc-labeled hexamethylpropyleneamine oxime (HM-PAO) in the brain was measured by single photon emission computed tomography (SPECT) and corrected for the nonlinearity caused by differences in net extraction. The "linearization" correction was based on a three compartment model, and it required a region of reference to normalize the SPECT image in terms of regional cerebral blood flow distribution. Two different regions of reference, the cerebellum and the whole brain, were tested. The uncorrected and corrected HM-PAO images were compared with cerebral blood flow (CBF) image measured by the C15O2 inhalation steady state method and positron emission tomography (PET). The relationship between uncorrected HM-PAO and PET-CBF showed a correlation coefficient of 0.85 but tended to saturate at high CBF values, whereas it was improved to 0.93 after the "linearization" correction. The whole-brain normalization worked just as well as normalization using the cerebellum. This study constitutes a validation of the "linearization" correction and it suggests that after linearization the HM-PAO image may be scaled to absolute CBF by employing a global hemispheric CBF value as measured by the nontomographic 133Xe clearance method.