Statistical parametric mapping (SPM) has been developed by Friston et al. to analyze focal changes in regional cerebral blood flow (rCBF) induced by brain activation. This automated and objective approach has the potential of being applied to single photon emission computed tomography (SPECT) image data sets. This study evaluated an automated analysis of N-isopropyl-p[123I]iodoamphetamine SPECT (123I-IMP) imaging in patients with dementia of Alzheimer's type (DAT) using statistical parametric mapping '95 (SPM'95). Eighteen patients with clinically diagnosed mild to severe DAT who were classified into two groups and ten normal control subjects were studied. The SPECT device used was a triple-headed rotation gamma camera. All images were transformed into the standard anatomical space of the stereotaxic brain atlas of Talairach and Tournoux, and then smoothed. Statistical analyses were made between severe DAT, mild DAT and normal control groups. SPM analysis of 123I-IMP SPECT images revealed that the parietotemporal association areas and posterior cingulate gyri of both cerebral hemispheres were significantly decreased in CBF in mild DAT group compared to normal controls. In comparison of DAT groups, the right hippocampal area and basal forebrain were significantly decreased in CBF in severe DAT group. In conclusion, these results obtained by SPM analysis of SPECT images suggested that the right hippocampal area and basal forebrain followed the parietotemporal association areas and posterior cingulate gyri in the reduction of the rCBF in patients with DAT.