The neurobiological mechanisms of both sleep and circadian regulation have been unraveled partly in the last decades. A network of brain structures, rather than a single locus, is involved in arousal state regulation, whereas the suprachiasmatic nucleus (SCN) has been recognized as a key structure for the regulation of circadian rhythms. Although most models of sleep regulation include a circadian component, the actual mechanism by which the circadian timing system promotes--in addition to homeostatic pressure--transitions between sleep and wakefulness remains to be elucidated. Little more can be stated presently than a probable involvement of neuronal projections and neurohumoral factors originating in the SCN. This paper reviews the relation among body temperature, arousal state, and the circadian timing system and proposes that the circadian temperature rhythm provides an additional signaling pathway for the circadian modulation of sleep and wakefulness. A review of the literature shows that increased brain temperature is associated with a type of neuronal activation typical of sleep in some structures (hypothalamus, basal forebrain), but typical of wakefulness in others (midbrain reticular formation, thalamus). Not only local temperature, but also skin temperature are related to the activation type in these structures. Warming of the skin is associated with an activation type typical of sleep in the midbrain reticular formation, hypothalamus, and cerebral cortex (CC). The decreasing part of the circadian rhythm in core temperature is mainly determined by heat loss from the skin of the extremities, which is associated with strongly increased skin temperature. As such, alterations in core and skin temperature over the day could modulate the neuronal activation state or "preparedness for sleep" in arousal-related brain structures. Body temperature may thus provide a third signaling pathway, in addition to synaptic and neurohumoral pathways, for the circadian modulation of sleep. A proposed model for the effects of body temperature on sleep appears to fit the available data better than previous hypotheses on the relation between temperature and sleep. Moreover, when the effects of age-related thermoregulatory alterations are introduced into the model, it provides an adequate description of age-related changes in sleep, including shallow sleep and awakening closer to the nocturnal core temperature minimum. Finally, the model indicates that appropriately timed direct (passive heating) or indirect (bright light, melatonin, physical activity) manipulation of the nocturnal profile of skin and core temperature may be beneficial to disturbed sleep in the elderly. Although such procedures could be viewed by researchers as merely masking a marker for the endogenous rhythm, they may in fact be crucial for sleep improvement in elderly subjects.