SUMMARYSeventy-three elderly patients (38 with Alzheimer-type dementia (ATD) and 35 with major depressive disorder) were followed up 2-5 years after an index admission during which a dexamethasone suppression test (DST) had been performed. Clinical state, cognitive function, neurological status and repeat DST were assessed where possible. The death rate was high in both groups (greater in the ATD group) but was not influenced by original DST status. Original DST status did not predict on survival time, development of physical illness or admission rate in either group, but DST non-suppression in the depressed group was associated with significant cognitive decline and abnormal neurological features. Possible mechanisms of the association are discussed.KEY wows-Alzheimer-type dementia, major depression, dexamethasone suppression test.Extensive studies have revealed that many depressed patients exhibit state-dependent hypothalamic-pituitary adrenal axis dysregulation with an increase in mean 24-hour plasma cortisol levels, a blunting of the normal diurnal rhythm of cortisol secretion and increased 24-hour urinary 17-hydroxy corticosteroid excretion (Carroll et al., 1981). More recently attention has focused on the observation that depressed patients show relative resistance to the suppressive effects of synthetic corticosteroids such as dexamethasone and a significant proportion of depressed patients fail to suppress on the dexamethasone suppression test (DST, Carroll et al., 1981). Initially it was suggested that the DST may distinguish between those elderly patients with depression and those with dementia (Carnes et al., 1983;Rudorfer and Clayton, 1982). However, recent studies have cast doubt on the DST as a diagnostic discriminator in elderly psychi- (1983) and McKeith (1984) found rates of DST non-suppression of51% and 58% respectively in Alzheimer-type dementia (ATD). In 1987 we reported a 47% rate of DST non-suppression in a group of ATD patients (Ferrier et al., 1988) although many fewer ATD patients than depressed patients exhibited basal hypercortisolaemia in our study.Therefore the DST has limited utility in routine diagnostic clinical practice in the psychiatry of old age. However, there are some outstanding issues to resolve. One of these is whether DST non-suppression identifies groups of demented andlor depressed patients with particular clinical characteristics and whether there are prognostic implications of DST non-suppression. This question is particularly important in view of the hypothesis put forward by Sapolsky (1987) that elevated glucocorticoids may induce selective hippocampal damage. Animal evidence indicates that the hippo-