Twelve outpatients with seasonal affective disorder (depression, winter type) were treated by 1 h of bright light exposure for five mornings. The intervention produced a significant reduction in depression scores, but no change was seen in the sleep electroencephalographic variables recorded after light treatment. Significant changes were seen, however, in ratings of subjective sleepiness. The acrophase of the circadian sleepiness rhythm was phase advanced, the mean level of the sleepiness rhythm was diminished, and the mean values of sleepiness scores were reduced at 8 and 10 a.m. This minimal influence of bright light on sleep structure is unlikely to explain the well-documented antidepressant effect.