Normal as well as disturbed sleep shows a considerable number of awakenings and sleep-wake transitions. This sleep fragmentation is quantified in terms number of arousals, awakenings, and the relative time of wakefulness during sleep related to the time spent in bed (sleep efficiency). In 48 healthy volunteers and 48 age matched patients with sleep disorders we evaluated the duration of sleep stages and wakefulness during sleep. We investigated the distribution of the durations in order to detect underlying statistical properties by applying single and double logarithmic scales. The distribution of sleep stages follows an exponential distribution. The different sleep stages have different time constants. In contrast to this the distribution of wake during sleep follows a power law. This was found for normal subjects and for patients. The patients had different time constants and different exponents in the power distribution. In order to check the general validity of these distributions we analyzed sleep wake transitions in scored sleep recordings from cats, rats and mice. There the same laws for the distribution were confirmed. This principally different regulation of sleep stages and wakefulness related to the distribution of durations can be only caused by different neuronal structures in the brain. Accordingly this analysis can be used to characterize sleep fragmentation species and in different sleep disorders. Lo C-C, Chou T, Penzel T, et al. PNAS 2004; 101: 17545–17548.