Circadian and sleep/wake dependent processes underlying variations in subjective alertness and cognitive performance were assessed in a constant routine protocol and in a protocol in which the sleep/wake cycle was uncoupled from the output of the endogenous circadian pacemaker. In the latter protocol, the contribution of a sleep/wake dependent process and a circadian process to alertness and performance were separated by folding the data at either the period of the sleep/wake cycle or at the period of the endogenous circadian body temperature rhythm. This analysis revealed that prior wakefulness within a range of 0-18 h significantly reduced alertness and performance and that the circadian rhythm of core body temperature paralleled the circadian rhythm of alertness and performance. During the first 16 h of the constant routine protocol, which coincided with the subjects' habitual period of wakefulness, alertness and performance remained at a stable level. The latter finding was explained by assuming that during our usual waking day the circadian system counteracts the detrimental effects of increasing duration of prior wakefulness.