In this paper, weinvestigate the underlying neuronal mechanism of Central Sleep Apnea (CSA) bystudying the sleep-wake dynamics as derived from hypnograms. We find that sleeparousals of CSA patients show a characteristic time scale (i.e., exponentialdistribution) in contrast to the scale-invariant (i.e., power-law) distributionthat has been reported for arousals in healthy sleep. We propose a mechanismfor this change in CSA wake bout distributions caused by the absence of thesleep restoring current at the transition from sleep to wake.