It is now well established that there are circadian (approximately 24 hour) rhythms in almost every physiological and psychological measure in man 1. These rhythms are not simply reactions to changes in the gross behavior or immediate environment of the subject, but stem from an internal (endogenous) timekeeping system. This ‘circadian system’ is self-sustaining in the absence of all time cues 2, and resistant to change when acute shifts in a person's routine are required 3,4. This resistance to change explains many of the adverse symptoms (e.g. malaise, insomnia, gastro-intestinal distress) complained of by those engaged in rotating shift work schedules, or suffering from ‘jet-lag’ 5,6. Such symptoms result from both an inappropriate phasing (time of peak) of circadian rhythms, and from a breakdown in the normal phase relationships between different rhythms as they adjust at different rates to the new routine. The latter is often referred to as ‘internal desynchronization’, and also occurs when different rhythms run at different periods. As an example of the importance of this kind of research, we discuss in this article a recent letter to Nature by Alain Reinberg and his associates 7 in which lability of the circadian temperature rhythm (particularly in its period) was found to be associated with a poor tolerance to rotating shift work.