Sleep is a complex, global and reversible behavioral state of all mammals, that is homeostatically regulated. Generally it is also defined as a rapidly reversible state of immobility and reduced sensory responsiveness. Still, there is no definition that has succeded in satisfying all aspects of sleep. The failure to define sleep as a single behavior lies in several facts: (1) sleep is not a homogenous state, but continuum of number of mixed states; (2) the control mechanisms of sleep are manifested at all levels of biological organization--from genes and intracellular mechanisms to the networks of neuronal populations within the central nervous system that control movement, arousal, autonomic functions, behavior and cognition; (3) the activity and interactions of these neurochemically greatly heterogenous neuronal populations are dependent of two biological rhythms--the circadian rhythm of wake/sleep and periodic cycles of NREM/REM sleep as two main sleep states. There are several levels of sleep control. The brain forebrain areas serve to control neuropsychology of dreaming; thalamo-cortical system controls NREM sleep rhythms, EEG activation and deactivation; hippocampo-cortical system controls memory consolidation; hypothalamic nuclei are the sources of circadian rhythm and sleep onset control; the control of periodic NREM/REM cycling is within the pons. The wake promoting neuronal populations are within the brainstem, midbrain, hypothalamus and basal forebrain. The main pontine wake-promoting centers are the noradrenergic neurons of locus coeruleus, the serotonergic neurons of dorsal raphe nucleus and the cholinerigic neurons of pedunculopontine tegmental nucleus and laterodorsal tegmental nucleus. The reciprocal connections and interactions of these neurons, and their opposite discharge pattern activity from wake to NREM and REM sleep have been the background of reciprocal interaction hypothesis of REM sleep generation. The wake-promoting neurons at the midbrain level are glutamatergic neurons of midbrain reticular formation, dopaminergic neurons of substantia nigra and ventral tegmental area. All pontine and midbrain wake-promoting cells project dorsally to activate thalamocortical system and ventrally to activate hypothalamo-cortical and basalo-cortical systems. There are also hypothalamic wake-promoting histaminegic neurons within the tuberomammilary nucleus of the posterior hypothalamus, and the hypocretinergic neurons of lateral hypothalamus. Cholinergic neurons of the basal forebrain beside control of wakefulness are included in control of many wake-promoting behaviors such as attention, sensory procession and learning. Recent molecular studies suggest the nucleus suprachiasmaticus as a wake-promotin area of the brain. The most important endogenously accumulated metabolite during wakefulness that initiates sleep is adenosine. Beside adenosine, the sleep-initiating factors are also GABA, glycine, PGD2, and cytokines IL-1beta and TNF-alpha. Also, releasing hormone of growth hormone (GHRH) increases depth and duration of NREM sleep. On the bases of "flip-flop" hypothesis of REM sleep control two GABA neuronal populations within the pons--REM-off area (ventrolateral periaqueductal grey matter and lateral pontine tegmentum), and REM-on area (sublaterodorsal nucleus and precoeruleus) are reciprocally interconnected and inhibit each other. REM-off area is activated by orexinergic neurons of lateral hypothalamus, and inhibited by GABAergic and galanin neurons of hypothalamic ventrolateral preoptic nucleus. Monoaminergic neurons of dorsal raphe nucleus and locus coeruleus are excitatory, while cholinergic neurons of PPT/LDT are inhibitory modulators of REM-off area of "flip-flop" switch. Although a lot of knowledge and progress has been accumulated in sleep research, we still do not have a meaningful explanation for the actual function of sleep. Sleep deprivation for 2-3 weeks in rats is fatal. Many experimental evidence suggest a role of REM sleep in brain development and maturation, synaptic homeostasis, memory and learning, but sleep in not universal state with the same underlying vital function in all species. Even within the species that do meet the behavioral definitons of sleep behavior (about 50 of 60 000 vertebrates tested) still is questionable do they sleep for the same reason.