Data from rodent studies indicate that cumulative stress exposure may accelerate senescence and offer a theory to explain differences in the rate of aging. Cumulative exposure to glucocorticoids causes hippocampal defects, resulting in an impairment of the ability to terminate glucocorticoid secretion at the end of stress and, therefore, in increased exposure to glucocorticoids which, in turn, further decreases the ability of the hypothalamo-pituitary-adrenal axis to recover from a challenge. However, the consensus emerging from reviews of human studies is that basal corticotropic function is unaffected by aging, suggesting that the negative interaction of stress and aging does not occur in man. In the present study, a total of 177 temporal profiles of plasma cortisol from 90 normal men and 87 women, aged 18-83 yr, were collected from 7 laboratories and reanalyzed. Twelve parameters quantifying mean levels, value and timing of morning maximum and nocturnal nadir, circadian rhythm amplitude, and start and end of quiescent period were calculated for each individual profile. In both men and women, mean cortisol levels increased by 20-50% between 20-80 yr of age. Premenopausal women had slightly lower mean levels than men in the same age range, primarily because of lower morning maxima. The level of the nocturnal nadir increased progressively with aging in both sexes. An age-related elevation in the morning acrophase occurred in women, but not in men. The diurnal rhythmicity of cortisol secretion was preserved in old age, but the relative amplitude was dampened, and the timing of the circadian elevation was advanced. We conclude that there are marked gender-specific effects of aging on the levels and diurnal variation of human adrenocorticotropic activity, consistent with the hypothesis of the "wear and tear" of lifelong exposure to stress. The alterations in circadian amplitude and phase could be involved in the etiology of sleep disorders in the elderly.