The objective of this study was to assess the diagnostic value for Alzheimer disease (AD) of single-photon-emission computed tomography (SPECT) and magnetic resonance imaging (MRI), separately and in combination. The study was part of a two-stage population-based study of mental functioning among noninstitutionalized 65-to 85-year-olds living in Amsterdam, The Netherlands. Participants (n = 51) were randomly selected within strata of cognitive function to obtain a sample of AD patients (n = 10) and clinically normal subjects (n = 41), of whom 22 displayed some cognitive impairment and fulfilled criteria for "minimal dementia" according to the Cambridge Examination for Mental Disorders of the Elderly. Coronal T1-weighted MRI was used to visualize the medial temporal lobe. Medial temporal lobe atrophy (MTA) was assessed qualitatively on a 0-4 scale. Regional cerebral blood flow on SPECT was assessed with the use of technetium 99m-HMPAO in three manually drawn regions of interest (frontal, parietal, and temporoparietal). Ratios were calculated by using the cerebellum as the reference area. MTA differed significantly between AD patients and clinically normal subjects (p = 0.0009), with sensitivity for AD of 70% and a specificity of 76%. None of the three SPECT ratios differed between normal and AD subjects. The temporoparietal/cerebellar ratio had a sensitivity of 30% and a specificity of 71% as a cutoff of 0.76. When both tests were positive the combined sensitivity was low (20%), but the false-positive rate was also very low (5%). A negative result on MRI or any SPECT ratio yielded a high specificity (93%-98%) but also a high false-negative rate (60-80%). Adding SPECT to MRI seems useful only if a diagnosis of AD is suspected clinically and confirmation is needed. When the clinical probability that AD is absent is high, normal results on either MRI or SPECT may confirm this notion. Given the fact that structural imaging should be performed in a clinical workup for dementia, using MRI only would be the most cost-effective approach.