In the first part of the study, a comparison of 19 patients with Alzheimer''s disease (AD) with 19 age-matched controls was performed to establish diagnostic criteria of AD with positron emission tomography and 18F-2-fluoro-2-deoxygIucose. Patients had a mean age of 60.6 ± 7.1 years, and a mean score of 14.5 ± 7.3 in Mini Mental Status examination, and of 4.9 ± 0.9 on the Global Deterioration Scale. They fulfilled current research criteria (NINCDS-ADRDA) for probable AD and had been screened rigorously to exclude other potentially dementing conditions, in particular cerebrovascular disease. Regional glucose metabolism was always abnormal in temporoparietal association areas, at least unilaterally, and in most patients also in fronto-lateral association areas. Regional metabolism relative to whole-brain metabolism was always normal in cerebellum, brain stem, and lentiform nucleus, and in most patients also in visual and sensorimotor cortex. A metabolic ratio of typically affected to typically unaffected regions was designed to represent this characteristic pattern. It provided a complete separation of AD patients from normals. In the second part of the study, the diagnostic power of this ratio was tested in an independent sample of 56 patients with a mean age of 59.9 ± 11.3 years. Eight of these patients had subjective memory complaints, but were found completely normal at neuropsychological and clinical examination. Twenty-two had cognitive deficits due to diseases other than AD, among them 7 with cerebrovascular disease, and 26 had probable AD. The classification by the metabolic ratio as AD or non-AD was correct in 85%, with a sensitivity of 92, and a specificity of 80%.