We studied the role of neuropsychology and SPECT imaging in the diagnosis and staging of dementia of Alzheimer type (DAT) in 33 patients with mild disease and 30 matched controls. Working, episodic, semantic and remote memory were assessed. For diagnosis, it was determined by logistic regression analysis that one of the memory tests (delayed verbal recall of the Doors and People Test) could correctly classify subjects as DAT or controls in 97% of cases. For staging, stepwise regression analysis using five of the memory tests could predict 70% of the variance in Mini Mental State Examination (MMSE) scores. The measures most useful for staging were tests of immediate recall, and tests of working, semantic and autobiographical memory. In a separate experiment, SPECT imaging on 31 of the above 33 patients and 24 different controls was used to address the issue of diagnosis and staging. 99Tc-HMPAO SPECT scans were analysed quantitatively to obtain measures of regional cerebral blood flow (rCBF). Logistic regression analysis showed that three of the SPECT regions of interest (left posterior temporal, right high frontal and right posterior temporal) could correctly classify subjects in 75% of cases. Of note was the fact that 39% of the DAT patients had normal SPECT scans. SPECT data were of limited use in modelling disease severity; only 38% of the variance in MMSE scores could be predicted from SPECT data. In addition, we found that the pattern of rCBF in DAT was much more heterogeneous than previously described.