Twelve subjects with winter depression were treated with equal quanta of green or white light in a randomized crossover study. Each treatment condition consisted of 2 hours of exposure each morning for 1 week, with at least 1 interventing week without treatment. Ratings made without knowledge of treatment condition were done before and after each condition. Both treatments significantly reduced depression ratings. White light was significantly more effective than green light in reducing endogenous symptoms, but not the "atypical" symptoms that are common features of winter depression. Furthermore, sequence of treatment conditions influenced antidepressant responses. Broad-band white light may be optimal for maximizing therapeutic response.