The present study tests the clinical hypothesis that aging impairs homeostatic adaptations of cortisol secretion to stress. To this end, we implemented a short-term 3.5-day fast as an ethically acceptable metabolic stressor in eight young (ages 18-35 yr) and eight older (ages 60-72 yr) healthy men. Volunteers were studied in randomly ordered fed vs. fasting sessions. To capture the more complex dynamics of cortisol's feedback control, blood was sampled every 10 min for 24 h for later RIA of serum cortisol concentrations and quantitation of the pulsatile, entropic, and 24-h rhythmic modes of cortisol release using deconvolution analysis, the approximate entropy statistic, and cosine regression, respectively. The stress of fasting elevated the mean (24-h) serum cortisol concentration equivalently in the two age cohorts [i.e. from 7.2 +/- 0.35 to 11.6 +/- 0.71 microg/dL in young men and from 7.7 +/- 0.39 to 12.6 +/- 0.59 microg/dL in older individuals (P < 10(-7))]. The rise in integrated cortisol output was driven mechanistically by selective augmentation of cortisol secretory burst mass (P = 0.002). The resultant daily (pulsatile) cortisol secretion rate increased significantly but equally in young (from 94 +/- 6.3 to 151 +/- 15 microg/dL x day) and older (from 85 +/- 5.4 to 145 +/- 7.3 microg/dL x day) volunteers (P < 10(-4)). Nutrient restriction also prompted a marked reduction in the quantifiable regularity of (univariate) cortisol release patterns in both cohorts (P < 10(-4)). However, older men showed loss of joint synchrony of cortisol and LH secretion even in the fed state, which failed to change with metabolic stress (P < 10(-6)). In addition, older individuals maintained a premature (early-day) cortisol elevation in the fed state and unexpectedly evolved an anomalous further cortisol phase advance of 99 +/- 16 min during fasting (P < 10(-5)). Caloric deprivation in aging men also disproportionately elevated the mesor of 24-h rhythmic cortisol release (P = 10(-7)) and elicited a greater increment in the mean day-night variation in cortisol secretory-burst mass (P < 0.01 vs. young controls). Lastly, short-term caloric depletion in older subjects paradoxically normalized their age-associated suppression of the 24-h rhythm in cortisol interburst intervals. In summary, acute metabolic stress in healthy aging men (compared with young individuals) unmasks distinct, albeit complex, disruption of cortisol homeostasis. These dynamic anomalies impact the feedback-dependent and time-sensitive coupling of pulsatile and 24-h rhythmic cortisol secretion. Nutrient-withdrawal stress in the older male heightens the cortisol phase disparity already evident in fed elderly individuals. Conversely, the stress of fasting in young men paradoxically reproduces selected features of the aging unstressed (fed) cortisol axis; viz., abrogation of joint cortisol-LH synchrony and suppression of the normal diurnal variation in cortisol burst frequency. Whether fasting would unveil analogous disruption of feedback-dependent control of the corticotropic axis in healthy aging women is not yet known.