Measurements of cerebral metabolism in patients with Alzheimer's disease (AD) using PET are artifactually depressed due to partial volume averaging of brain tissue activity with enlarged CSF spaces. To investigate the effects of correction for the expansion of CSF spaces on regional metabolic measures, as well as the correlations between neuropsychological test results and resting cerebral metabolism before and after partial volume correction, we applied an MRI-based method of partial volume correction to 18F-fluorodeoxyglucose (FDG)-PET data from eight patients diagnosed with probable AD and ten healthy elderly individuals. Before correction, the AD group had significantly lower cortex-to-cerebellum ratios in the posterior temporal, parietal, and frontal lobes in comparison to the control subjects. Partial volume correction of PET data resulted in 19 to 49% increases in regional activity in the AD group and 16 to 38% increases in the control group. The patients' persistence of significant hypometabolism in the frontal, posterior temporal, and parietal regions after partial volume correction suggests that a true reduction in regional cerebral glucose metabolism occurs in AD, even though its magnitude is a result of both metabolic reductions and the effects of atrophy. Partial volume correction of PET data in the AD group had a significant impact on the correlations between regional glucose metabolism and neuropsychological performance. These findings suggest that accounting for differential extent and distribution of cerebral atrophy in patients with AD and in healthy individuals may potentially improve our ability to interpret specific cognitive dysfunction in the context of the functional imaging data.