Oral temperatures have been measured regularly in a group of institutionalized psychiatric patients. Each subject had his temperature measured on at least two occasions, once when Greenwich Mean Time was in operation (November-December) and at least once when British Summer Time was in use (July-October). There was no difference between the mesors and amplitudes of the rhythms measured at different times of the year. When all results were expressed on Greenwich Mean Time, the acrophase measured in the summer or autumn tended to be later than that measured in the same subject in winter. The difference in acrophase could not be attributed to differences in routine of the subjects and suggests that the time of sunset has an effect upon temperature rhythms. The implication of this result in terms of the relative importance of social factors and light as synchronizers is discussed.