Critically ill patients have elevated plasma cortisol concentrations, in proportion to illness severity. This was traditionally attributed exclusively to a central activation of the hypothalamus-pituitary axis. However, low rather than high plasma ACTH concentrations have been reported in critically ill patients, with loss of diurnal ACTH and cortisol rhythm. Low ACTH together with high cortisol is referred to as "ACTH-cortisol dissociation." Although cortisol production is somewhat increased with inflammation, a reduced cortisol breakdown explains to a larger extent the hypercortisolism during critical illness. Inflammation-driven decrease in cortisol binding proteins further increase the active free cortisol fraction. Several drugs administered to ICU patients suppress plasma cortisol in a dose-dependent manner. Sustained low circulating ACTH might contribute to adrenal atrophy and dysfunction in the prolonged phase of critical illness. In the acute phase of sepsis or septic shock, a condition referred to as "relative adrenal insufficiency" has been suggested to ensue from glucocorticoid resistance and insufficiently elevated circulating cortisol to overcome such resistance, with pathological changes possibly occurring at every level of the HPA axis. However, it remains highly controversial whether tissue-specific glucocorticoid resistance is adaptive or maladaptive, how to diagnose "relative" adrenal insufficiency, and how it should be treated. Large RCTs, investigating the effect of 200 mg/d hydrocortisone treatment for sepsis or septic shock have shown conflicting, mainly negative, results. Not taking into account the reduced cortisol breakdown, which increases the risk of overdosing hydrocortisone, might have played a role. Further research on diagnostic, therapeutic and dosing aspects is urgently warranted. Compr Physiol 8:283-298, 2018.