It has been shown that long-term estrogen treatment in gonadectomized female mice increases anxiety levels. On the other hand, a recent study has reported that estrogen may down-regulate the levels of anxiety by acting through estrogen receptor (ER) beta. In the present study, we investigated the role of ER-beta in the regulation of anxiety levels in female mice after long-term estrogen treatment. Gonadectomized ER-beta knockout (betaERKO) female mice and their wild type (betaWT) littermates were implanted several different doses (experiment 1: 2.0 microg/day, experiment 2: 1.0, 0.4, 0.2 or 0.1 microg/day) of an estradiol benzoate (EB) or placebo pellet. Ten days after pellet implant, behavioral tests commenced to measure the anxiety levels (experiment 1: light-dark transition test (LDT), experiment 2: LDT, elevated plus maze test (EPM) and social investigation test (SIT)). We found that, at higher-doses, long-term treatment of EB had anxiogenic effects in both betaWT and betaERKO mice as indicated by a decrease of the time spent in the light side and the number of transitions between two sides during LDT. In contrast, several behavioral measurements indicated that the lower-doses treatment of EB might reduce the anxiety levels possibly through ER-beta. Particularly, the anxiolytic effects of EB in the SIT were more pronounced in betaWT mice than betaERKO mice. Together, the findings in the present study suggest that estrogen may have both anxiolytic and anxiogenic effects in female mice, and that ER-beta gene disruption did not affect anxiogenic regulation by estrogen in female mice, but partially affected anxiolytic regulation.