Abnormalities in circadian rhythms of core body temperature have been reported previously in depressed patients. In this study, we compared the temperature rhythms of 10 depressed seasonal affective disorder (SAD) patients with winter depression with those of 12 normal controls and evaluated the effects of bright light on temperature in SAD. Unlike previous studies of depressed patients, the temperature curves of the patients and normal controls during the off-light condition were nearly identical. We found a significant difference in amplitude between the patients in the untreated and light-treated conditions. Although there was no systematic difference in circadian phase across groups or treatment conditions, we present preliminary evidence that suggests that phase-typed subgroups may be present in the population distinguished by their treatment responses.