Positron emission tomography (PET) has been promoted as a means of improving the diagnosis of Alzheimer's disease (AD), but the evidence to support its incremental value is unclear. To assess the evidence regarding the use of PET in the clinical evaluation of AD, a systematic review of the English-language literature indexed in MEDLINE (1975–January 2001), the Cochrane Library (issue 4, 2000), and health technology assessment (HTA) reports was conducted. Articles identified by this review process were graded for methodological and reporting quality using a standardized grading scheme. Sixteen original articles and seven HTA reports were identified. In general, the articles addressed: using PET to differentiate AD from normal aging or non-Alzheimer's dementias, PET imaging compared with single positron emission computed tomography imaging, using PET to predict the progression of dementia, and agreement and reliability in the interpretation of PET images. Serious problems with study design and methodology in all articles were identified. Previous HTA reports have generally recommended that PET not be used in the clinical evaluation of dementia. In conclusion, there is little evidence to support the addition of PET to the routine clinical evaluation of patients with suspected or established dementia. Suggestions for future research in this area are offered.