Although it is widely believed that benzodiazepines reduce anxiety through positive allosteric modulation of the GABA(A)-chloride channel complex, this is not the only mechanism through which agents of this class can modify CNS function. Furthermore, a significant number of reports of apparent flumazenil blockade of diazepam anxiolysis in animal models have paid limited attention to possible intrinsic behavioral actions of the antagonist per se. In the present study, ethological methods were employed to assess in detail the effects of diazepam, flumazenil, and their combination on the behavior of male DBA/2 mice in the elevated plus-maze paradigm. In two experiments, diazepam (1.5 mg/kg) alone reduced open-arm avoidance and increased head dipping, whereas flumazenil (10-40 mg/kg) alone was without significant behavioral effect. However, with the sore exception of head dipping, prior administration of flumazenil (10 and 40 mg/kg) failed to block the behavioral effects of diazepam under present test conditions. These findings imply that the anxiolytic effects of diazepam in the mouse plus-maze are not mediated through flumazenil-sensitive benzodiazepine receptors and that alternate mechanisms must be considered. (C) 1999 Elsevier Science Inc.