The effect of caffeine and coffee on the sleeping time induced by barbital was tested in rats. It was found that caffeine pretreatment shortened barbital-induced sleeping time and that coffee had an effect in direct proportion to its caffeine content; decaffeinated coffee was without effect. It was hypothesized that the shortening was due to an interaction at the brain level. This hypothesis was tested using hexobarbital which is metabolized by the liver whereas barbital is only negligibly so metabolized. Pretreatment with caffeine was found to shorten the hexobarbital sleeping time even though the brain concentration of the latter was actually elevated. It was also shown that caffeine pretreatment was much less effective than phenobarbital pretreatment in stimulating the enzyme systems in the liver which metabolize hexobarbital, aminopyrine and ethylmorphine (in vitro). Consequently, it is concluded that caffeine stimulation of liver metabolism is a relatively minor factor in the shortening of barbiturate sleeping time by caffeine in comparison to the brain interaction of these two drugs.