Experiments have been performed in an isolation unit to investigate the effects of abnormal sleep-waking schedules upon circadian rhythms of renal excretion and deep-body temperature. In confirmation of previous work, nychthemeral rhythms appeared to be 'anchored' to a 24 h period if 4 h sleep was taken regularly each day, even though another 4 h was taken irregularly. The endogenous components were investigated by assessing circadian rhythmicity under constant routine conditions, that is, when rhythmic influences in the environment and sleep-waking pattern had been minimized. Analysis of the constant routine data indicated the presence of a rhythmic component which had been stabilized to a period of 24 h by the 'anchor sleep'. In addition, a delayed component was also present. The starting time of the constant routines produced a direct effect upon the rhythms, which was presumed to result from removing the 'masking' effect that sleep normally exerts upon rhythms. There was some evidence that the relative importance of the masking effect and the delayed component depended upon the variable under consideration. The implications of these findings, in terms of the effects of anchor sleep, the presence of more than one internal clock and the usefulness of constant routines, are discussed.