Ten patients with seasonal affective disorder received the following treatments for 5 days each: (a) artificial daylight (2500 lux) from 20.00 to 23.00 and from 07.00 to 10.00 hours; (b) red light (300 lux) from 20.00 to 23.00 and from 07.00 to 10.00 hours; (c) artificial daylight (2500 lux) from 22.00 to 23.00 and from 07.00 to 08.00 hours. The antidepressant effect of treatment (a) was superior to that of treatment (b), suggesting that the effect of light treatment in winter depression is more than that of a placebo. The antidepressant effect of treatment (a) was superior to that of treatment (c), although these two treatments equally suppressed plasma melatonin concentrations. Consequently, in these patients there is a dissociation between the effect of light treatment on melatonin and the reduction of depression ratings.