To examine the utility of the dexamethasone suppression test (DST) in the differential diagnosis of depression in elderly demented patients, we reviewed the literature and focused on four components of this question: (1) cortisol nonsuppression rates in dementia; (2) cortisol nonsuppression and dementia severity; (3) cortisol nonsuppression in demented versus depressed patients; and (4) cortisol nonsuppression following antidepressant treatment. A combined analysis of 27 articles showed cortisol nonsuppression in 60% of patients with concurrent dementia and depression, in 47% of patients with depression only, in 41% of patients with dementia only, in 46% of patients with multi-infarct dementia, in 36% of patients with primary degenerative dementia, and in 10% of controls. The abnormal DST rate in demented patients was not significantly different from the abnormal DST rate in depressed patients. Eight of 12 studies (67%) did not find a significant relationship between DST results and dementia severity in dementia patients without depression. Twelve of 13 studies (92%) did not find a relationship between age and DST outcome. The data we reviewed do not support the use of the DST in discriminating between depression and dementia or between dementia subtypes. (J Geriatr Psychiatry Neurol 1990;3:124-138).