The association between fatigue and reduced activity in the hypothalamo-pituitary-adrenal (HPA) axis has been described. However the temporal association between fatigue and HPA activity is under debate. We examine whether alterations in cortisol secretion play a role in the development of fatigue or whether changes occur later as a consequence of fatigue in a longitudinal cohort study of 4299 community dwelling adults (mean age 61). Cortisol secretion was measured from saliva samples collected waking, waking + 0.5, 2.5, 8, 12 h and bedtime at phase 7 (2003-2004) of the Whitehall II study. Fatigue was measured at phase 6 (2001), phase 7 and phase 8 (2006) of the Whitehall II study. Three elements of secretion were examined: waking cortisol, the cortisol awakening response and diurnal slope in cortisol secretion. Fatigue was determined using the vitality sub-scale of the Short Form-36. A wide variety of co-variates were measured. We find that fatigue measured at phase 6 was not associated with cortisol secretion at phase 7. At phase 7, low waking cortisol levels and a flat slope in diurnal cortisol secretion were associated with fatigue independently of co-variates. In participants low or free of fatigue at phase 7 low waking cortisol and flatter slope in cortisol secretion were associated with new-onset fatigue at phase 8 (for example, odds ratio for lowest vs. highest tertile of waking cortisol 1.50; 95% confidence intervals, 1.08, 2.09 after adjusting for all co-variates). In conclusion, we find that low waking salivary cortisol and a flat slope in cortisol secretion is associated with fatigue. Cortisol is also associated with future onset of fatigue suggesting that changes in cortisol secretion are etiologic or occur early in the genesis of fatigue.