This study evaluated the effect of using two different reference regions in the quantification of single-photon emission tomography (SPET). SPET scans of 30 patients with Alzheimer's disease (AD) and 30 patients with frontotemporal dementia were compared with the scans of ten age-matched controls. Regions of interest (ROIs) were defined on transaxial slices by a semi-automatic method. Regional cerebral blood flow indices (rCBFi) in each ROI were determined by normalizing the count densities to both cerebellar and occipital cortex reference regions. Mean rCBFi for each ROI were calculated for the patient and control groups and significant group differences determined. The number and topographical distribution of ROIs with significant group differences varied depending upon the choice of reference region. The magnitude of these differences was greatest when the cerebellum was used as the reference region. The disparity between results obtained with the two reference regions was most apparent in the AD group. The reasons for these differences are discussed and we conclude that the cerebellum is the more appropriate choice of reference region in the quantification of SPET in primary degenerative dementia.