Experiments investigating timing behaviour in humans under conditions where body temperature was raised or (much more rarely) lowered, dating from 1927 to 1993, were reviewed. These tested the hypothesis that humans possess a temperature-sensitive chemical or biological internal clock. Most studies used conditions in which subjects produced or estimated durations less than 100 sec long, probably using chronometric counting, but other experimental paradigms were sometimes employed. Data from each study were expressed in a uniform fashion, as plots of changes in the rate of subjective time (estimated from changes in timing behaviour) against changes in body temperature. In almost all cases, rate of subjective time increased when body temperature increased above normal, and decreased when body temperature was lowered below normal, although observations of the latter type were rare. The data also suggested a parametric effect of body temperature, with higher temperatures generally producing faster subjective time. Some possible mechanisms for the effects obtained were discussed, with the most promising explanation probably being that the temperature manipulation produces changes in arousal.