Hypercortisolism in depression has been extensively studied during the last three decades. The main hypothesis regarding origin and clinical relevance of this phenomenon, however, has changed significantly. Up to the mid-seventies hypercortisolism was conceived as consequence of stress modified by the degree of unconscious defense mechanisms in different forms of depressive or non-depressive psychiatric disorders. At the end of the seventies this point of view changed considerably. Hypercortisolism was regarded as a biological statemarker of the endogenous subtype of depression with clinical differential-diagnostic relevance. An abnormal dexamethasone suppression test (DST) was assumed to be the best indication of increased activation of the cortisol system. These conclusions turned out to be wrong. DST results are not specific for melancholia and the test seems to be of limited value for measuring the function of the HPA-axis. Intervening variables, such as weight loss, drug and alcohol withdrawal or situational stress, influence the test results significantly, independent of the nosological classification. Additionally, interindividual differences in the susceptibility of the HPA-axis may decisively influence the the activation of the HPA-axis as well in healthy subjects under stress as in psychiatric patients.