To determine whether a cumulative sleep debt (in a range commonly experienced) would result in cumulative changes in measures of waking neurobehavioral alertness, 16 healthy young adults had their sleep restricted 33% below habitual sleep duration, to an average 4.98 hours per night [standard deviation (SD) = 0.57] for seven consecutive nights. Subjects slept in the laboratory, and sleep and waking were monitored by staff and actigraphy. Three times each day (1000, 1600, and 2200 hours) subjects were assessed for subjective sleepiness (SSS) and mood (POMS) and were evaluated on a brief performance battery that included psychomotor vigilance (PVT), probed memory (PRM), and serial-addition testing, Once each day they completed a series of visual analog scales (VAS) and reported sleepiness and somatic and cognitive/emotional problems. Sleep restriction resulted in statistically robust cumulative effects on waking functions. SSS ratings, subscale scores for fatigue, confusion, tension, and total mood disturbance from the POMS and VAS ratings of mental exhaustion and stress were evaluated across days of restricted sleep (p = 0.009 to p = 0.0001). PVT performance parameters, including the frequency and duration of lapses, were also significantly increased by restriction (p = 0.018 to p = 0.0001). Significant time-of-day effects were evident in SSS and PVT data, but time-of-day did not interact with the effects of sleep restriction across days. The temporal profiles of cumulative changes in neurobehavioral measures of alertness as a function of sleep restriction were generally consistent. Subjective changes tended to precede performance changes by 1 day, but overall changes in both classes of measure were greatest during the first 2 days (P1, P2) and last 2 days (P6, P7) of sleep restriction. Data from subsets of subjects also showed: 1) that significant decreases in the MSLT occurred during sleep restriction, 2) that the elevated sleepiness and performance deficits continued beyond day 7 of restriction, and 3) that recovery from these deficits appeared to require two full nights of sleep. The cumulative increase in performance lapses across days of sleep restriction correlated closely with MSLT results (r = -0.95) from an earlier comparable experiment by Carskadon and Dement (1). These findings suggest that cumulative nocturnal sleep debt had a dynamic and escalating analog in cumulative daytime sleepiness and that asymptotic or steady-state sleepiness was not achieved in response to sleep restriction.