The effect of bright light on cortisol and the relationship between melatonin and cortisol were studied in 63 depressed patients (42 patients with a seasonal pattern and 21 patients with a non-seasonal pattern). The patients were matched for age, time of treatment and severity of depression. Before and after light treatment the severity of the depression was rated with the Comprehensive Psychopathological Rating Scale (23 items) and the Hamilton Depression Rating scale (18 items), and serum cortisol and melatonin were drawn at nine time-points between 20.00 and 08.00 hours. Two hours of light treatment (350 cd m-2) was given daily for 10 days either in the morning (06.00-08.00 hours) or in the evening (18.00-20.00 hours). As reported earlier, patients with a seasonal pattern improved significantly more than patients with a non-seasonal pattern of depression, and no significant differences were found between the treatment efficacy of morning compared to evening light. A cosinor analysis showed that the cortisol batyphase was significantly advanced by morning light, but was not delayed by evening light. A delay in batyphase cortisol showed a weak significant correlation with a decrease in the absolute and relative sum of scores. The batyphase of cortisol occurred approximately 3 h earlier than the acrophase of melatonin. Of the changes in the melatonin acrophase 43% were reflected in a change of cortisol batyphase, indicating a hierarchical relationship with melatonin as the co-ordinating hormone transducing part of the information of the external light to the phase position of cortisol. No significant differences between patients with a seasonal or a non-seasonal pattern were seen in mesor, amplitude or batyphase of cortisol before treatment, and no significant changes in mesor or amplitude were seen as a result of light treatment.