This study aimed to describe quantitatively some changes in sleep behavior. During 70 consecutive nights, 28 women and 3 men, 30 to 40 years of age and presumably clinically healthy, recorded the time of each awakening. Times of falling asleep were estimated from markings at 10-min. intervals from the times of lying down to sleep as an indication that the subject had not yet fallen asleep. Sleep duration and an index of effective sleep derived therefrom were analyzed by rhythmometric methods. On a group basis, anticipated components with periods of 1 and 0.5 wk., synchronized with the social schedule, were detected with statistical significance. Until long-term polysomnographic monitoring can readily cover the week automatically rather than only one or a few daily sleep spans, the self-monitoring of sleep behavior, yielding the circaseptan endpoints derived herein, may serve as a cost-effective tool in sleep research. By virtue of their relative simplicity, they could be part of a protocol designed to assess pharmacologic or nonpharmacologic interventions of sleep disturbance aimed at restoring undisturbed sleep.