The large night-time growth hormone (GH) pulses are associated with slow wave sleep (SWS) phases in humans and a similar association has been found also in some other mammalian species. According to previous pharmacological studies an administration of growth hormone-releasing hormone (GHRH) promotes sleep, especially SWS, in humans and laboratory animals. Similar studies in rats have shown that somatostatin promotes REM sleep. The neuropeptide galanin increases GH-secretion and induces sleep in humans. The cells synthetising GHRH, somatostatin and galanin are mainly located in the hypothalamic areas. In order to study the possible endogenous role of GHRH, somatostatin and galanin in sleep regulation we measured the effect of REM and total sleep deprivation on the expression of mRNA of these neuropeptides in the rat hypothalamus by in situ hybridization. We also measured the effect of galanin, somatostatin and somatostatin antagonist on sleep by a polygraphic method in the course of which these peptides were injected intracerebroventricularly (i.c.v.) or locally into the locus coeruleus of rats. We observed that selective REM sleep deprivation increased the amount of somatostatin mRNA in the arcuate nucleus and decreased GHRH mRNA in the paraventricular nucleus. Combined loss of REM sleep and SWS achieved by total sleep deprivation increased both somatostatin and GHRH mRNA in these hypothalamic nuclei. Because the paraventricular (GHRH) and arcuate (somatostatin) nuclei are not the main hypophysiotropic nuclei of GHRH and somatostatin, control of sleep and wakefulness and GH regulatory functions of these neuropeptides are presumably performed by different cell populations in the hypothalamus. REM sleep deprivation increased the amount of galanin mRNA in the hypothalamus, while total sleep deprivation had no effects. I.c.v. or locus coeruleus microinjections of galanin did not have effects on sleep. These findings suggest that galanin may mediate some effects of REM sleep deprivation in the brain, but does not regulate REM sleep by itself. I.c.v. or local injection of somatostatin did not affect daytime sleep in rats, but a somatostatin antagonist reduced REM sleep when injected i.c.v. or locally into the locus coeruleus. I.c.v. injection of somatostatin antagonist also reduced REM sleep when REM sleep propensity was elevated after REM sleep deprivation. The present findings support the hypothesis that endogenous GHRH promotes SWS  while somatostatin promotes REM sleep. We hypothesize that the locus coeruleus may partly mediate the effects of somatostatin on REM sleep.