We previously showed that the extract of Japanese angelica root (JAR-E) reversed the decrease in pentobarbital (PB) sleep induced by isolation stress and yohimbine and methoxamine, stimulants of central noradrenergic systems, in mice. Here, we tested the effects of several fractions from JAR-E and ligustilide and butylidenephthalide, phthalide components of JAR-E, on PB sleep in isolated mice to elucidate the mechanism of the action of JAR-E. Methanol-soluble (Met-S) and -insoluble (Met-IS) fractions were obtained from JAR-E. Methylenechloride-soluble (MC-S) and -insoluble fractions (MC-IS) were prepared from Met-S. MCS (11.4-76 mg/kg, p.o.) reversed the isolation stress-induced decrease in PB sleep, but neither Met-IS (0.8-2.4 g/kg, p.o.) nor MC-IS (0.7-2 g/kg, p.o.) had the same effect. The i.p. administration of MC-S exhibited a similar activity to that observed after the p.o. administration of the same fraction. Ligustilide (5-20 mg/kg, i.p.) and butylidenephthalide (10-30 mg/kg, i.p.) reversed PB sleep decrease in isolated mice. Both components (20 mg/kg, i.p.) attenuated the suppressive effects of yohimbine (30 nmol, i.c.v.), methoxamine (200 nmol, i.c.v.) and a benzodiazepine inverse agonist FG7142 (10 mg/kg, i.p.) on PB sleep in group-housed mice. These results suggest the contribution of ligustilide and butylidenephthalide to the effect of JAR-E on PB sleep in isolated mice, and implicate central noradrenergic and/or GABA, systems in the effects of these components.