We review the literature on whether exercise is a synchroniser of human circadian rhythms, and highlight the specific methodological problems that are associated with this topic. In applied research, exercise has been investigated as a treatment for jet lag and shift-work problems. In these studies, there have been difficulties in controlling the characteristics of the exercise bout, the athletic status of research participants and exposure to other confounding synchronisers. Therefore, it is unclear at present whether exercise can help mitigate the problems associated with transmeridian travel and shift-work. In laboratory-based experiments, participants have exercised at various times of the day and the change in phase of various circadian rhythms has been measured. Although it is difficult to control for the considerable masking effects of exercise on these estimates of circadian timing, it is clear that nocturnal exercise can induce phase delays in the onset of melatonin. Reports of exercise-induced phase advances of the melatonin rhythm are rarer, as are any phase-shifting effects at all on the body temperature rhythm. In practical terms, the substantial levels of activity needed to obtain phase shifts may not be attainable by the majority of people. In mechanistic terms, the lack of agreement with the phase-shifting effects of bright light suggests that exercise is not exerting its effects via photic entrainment pathways. An alternative explanation may involve exercise-induced hyperthermia. Moreover, it is perplexing why exercise should have a different phase response curve to light, given that humans are diurnally active.