In winter, depressed patients with seasonal affective disorder respond to treatment with bright artificial light (phototherapy). The authors found that the antidepressant effects of phototherapy were much greater for 10 patients when light was applied to the eyes than when it was applied to the skin, suggesting that its effects may be mediated by the eyes. The identification of a probable anatomical route of entry is clinically relevant and an important clue for further investigations of the mechanism of phototherapy. However, patients' expectations nearly always predicted the outcome, leaving open the possibility that expectations were responsible for their responses.