Plasma cortisol concentration was measured at 20 min intervals from 3 p.m. (1500 hrs) to 6 p.m. (1800 hrs) in 26 hospitalized patients classified according to the Newcastle Index as endogenously depressed (n = 16) or non-endogenously depressed (n = 10). When examined in depressed state, before treatment, maximum, mean and range of plasma cortisol concentration in this time interval was significantly higher in the endogenously depressed patients than in the non-endogenously depressed patients (p less than 0.01-0.02). The diagnostic identification of endogenous depression on the basis of these cortisol concentration measurements was at least as good as that reported by others using post-dexamethasone cortisol levels. The plasma cortisol levels (maximum, mean) and fluctuations (range) correlated significantly with the degree of depression (Hamilton Depression Scale), and differences in severity of depression could explain most of the differences in cortisol levels between the two diagnostic groups. Nine patients were reexamined after 3-12 months in a non-depressed state, and all unipolar endogenously depressed patients (n = 6) then had clearly reduced cortisol levels and fluctuations.