The regulation of ketogenesis by the hepatic nerves was investigated in the rat liver perfused in situ. Electrical stimulation of the hepatic nerves around the portal vein and the hepatic artery caused a reduction of basal ketogenesis owing to a decrease in acetoacetate release to 30% with essentially no change in 3-hydroxybutyrate release. At the same time, as observed before [Hartmann et al. (1982) Eur. J. Biochem. 123, 521-526], nerve stimulation increased glucose output, shifted lactate uptake to output and decreased perfusion flow. Ketogenesis from oleate, which enters the mitochondria via the carnitine system, was also lowered after nerve stimulation owing to a decrease of acetoacetate release to 30% with no alteration in 3-hydroxybutyrate release. Ketogenesis from octanoate, which enters the mitochondria independently of the carnitine system, was decreased after nerve stimulation as a result of a drastic decrease of acetoacetate output to 15% and a less pronounced decrease of 3-hydroxybutyrate release to 65%. Noradrenaline mimicked the metabolic nerve effects on ketogenesis only at the highly unphysiological concentration of 0.1 microM under basal conditions and in the presence of oleate as well as partly in the presence of octanoate. It was essentially not effective at a concentration of 0.01 microM, which might be reached in the sinusoids owing to overflow from the hepatic vasculature. Sodium nitroprusside prevented the hemodynamic changes after nerve stimulation; it did not affect the nerve-dependent reduction of ketogenesis under basal conditions and in the presence of oleate, yet it diminished the nerve effect on octanoate-dependent ketogenesis. Phentolamine clearly reduced the metabolic and hemodynamic nerve effects, while propranolol was without effect. The present data suggest that hepatic ketogenesis was inhibited by stimulation of alpha-sympathetic liver nerves directly rather than indirectly via hemodynamic changes or noradrenaline overflow from the vessels and that the site of regulation should be mainly intramitochondrial.