Prior studies of patients with dementia have found similar qualitative patterns of cerebral glucose utilization with [18F]2-fluoro-2-deoxyglucose (FDG) PET and of putative muscarinic receptor activity with [123I]3-quinuclidinyl-4-iodobenzilate (IQNB). This raised doubts about whether receptor binding determines IQNB distribution and whether clinical information in IQNB scans is unique. To compare the methods directly, 4 normal volunteers and 7 patients with dementia underwent FDG PET and high-resolution IQNB SPECT scans. In normal subjects, relative regional activity from the paired scans was only weakly correlated (r = 0.29). Some regions (e.g., thalamus, frontal cortex) showed a clear disassociation of activity. In demented patients, IQNB scans tended to show larger defects than FDG scans, although one focal defect appeared only with PET. Results suggest that IQNB SPECT data are not primarily related to general physiological activity or regional cerebral blood flow and are not explained by attenuation or volume-averaging artifacts. Further studies should investigate whether IQNB scanning is a more sensitive in vivo measure of the extent of Alzheimer's disease than is FDG PET.