Twelve patients with seasonal affective disorder (SAD) and eleven normal controls were exposed to 2000 lux and 300 lux of artificial full-spectrum light on consecutive nights during the winter. Suppression of melatonin secretion under the two light intensities was measured and the difference between their effects was taken as a measure of light sensitivity. The test was repeated in summer in both groups, when the SAD subjects were well. The SAD but not the normal group showed a significant seasonal variation in sensitivity to light. There was evidence of supersensitivity in the winter but also of subsensitivity to light in the summer.