This study was designed to test whether or not flumazenil (1 mg IV) would change levels of anxiety induced by simulation of public speaking (PSS). Forty normal volunteers were randomly allocated in a parallel groups design to treatment with flumazenil or placebo (double-blind) and one of two stress conditions (PSS/control task). PSS increased anxiety and flumazenil antagonized this effect. The anxiolytic activity of flumazenil was particularly strong on anticipatory anxiety measures. The results were discussed in terms of the involvement of endogenous benzodiazepine-type ligands in the regulation of the human response to stress.