BACKGROUND: Cortisol hypersecretion is characteristic of acute clinical depression, but little is known in fully recovered, non-treated elderly persons with a lifetime history of depression. This study was designed to examine patterns of diurnal cycle of cortisol in an elderly cohort without current depression or treatment for depression according to whether the person has or has not experienced a previous episode of depression or co-morbid depression with anxiety. METHODS: Cortisol secretion was evaluated in 162 community-dwelling elderly on a stressful and a non-stressful day (basal level). Past depression and anxiety disorders were assessed using a standardized psychiatric examination based on DSM-IV criteria (the Mini International Neuropsychiatric Interview). RESULTS: Antidepressant-free persons with a history of non-co-morbid major depression (6.8% of the sample) showed basal cortisol hypersecretion compared to those with depression and anxiety (8.6%) or controls. Several hours after exposure to a stressful situation, controls showed a sustained increase in cortisol secretion, which was not observed in persons with a history of depression. Persons with a history of depression with anxiety showed a similar cortisol secretion at baseline to controls but a heightened response to stressful situation; a pattern comparable to that observed in subjects with pure anxiety disorders (16.7%). CONCLUSION: An abnormal HPA response persists even after effective treatment for depression. A history of co-morbid depression and anxiety gives rise to changes characteristic of anxiety alone. Our findings suggest that cortisol abnormalities may be trait markers for vulnerability to depression and for the differentiation of depression and depression with co-morbid anxiety.