Anxiolytic properties of melatonin in rodents had usually been examined in behavioral tests based on stressful situations, i.e., in animal models of "state" anxiety. However, no study reports effects of melatonin on emotionality of rodents submitted to situations devoid of stressful components as in the free-exploratory test, which gives to animals the opportunity to choose freely between familiar and unfamiliar places. This procedure has been proposed as a method for measuring an endogenous form of anxiety called "trait" anxiety. The present study first investigated the effects of melatonin on neophobic responses of male C57BL/6, C3H/He, and BALB/c mice submitted to a free-exploratory test. Results demonstrated that melatonin had no effect in C57BL/6 mice that presented very low neophobic responses, whereas it was effective in reducing neophobia of BALB/c and C3H/He mice that presented, respectively, strong and intermediate avoidance responses towards unfamiliarity. Indeed, mice of both latter strains treated with melatonin made fewer attempts to enter into the unfamiliar compartment, exhibited a lower latency of the first entry into the unfamiliar places, and spent more time in them. Thus, melatonin appeared to be equally effective in reducing "trait" anxiety in both BALB/c and C3H/He mice. Moreover, flumazenil was able to counteract, in a dose-dependent manner, the anxiolytic activity of melatonin in BALB/c, suggesting involvement of central GABAergic system in the pharmacological effects of melatonin.