A hypothesis is proposed which suggests that the requirements imposed by waking on a day to day basis dictates the way organisms sleep. This generates 'somnoprints' which are characteristic patterns of sleep directly related to events of waking. REM sleep is suggested to be the phase which contends with waking requirements by manifesting levels of excitability which have the dual function of generating the phase and maintaining it for period durations which are determined by the functional requirements imposed by the moment. Since the brain cannot maintain long periods of excitation, it always alternates with SWS for homeostatic reasons. The hypothesis includes the justification for the diversity of chemical signals regulating sleep, by suggesting that waking activities generate different needs which in turn generate different chemical signals and activate different set of neuronal nets who increase their excitability levels to maintain sleep through varying pathways.