Eight experiments supported the hypotheses that reflexive testosterone release by male mice during sexual encounters reduces male anxiety (operationally defined in terms of behavior on an elevated plus-maze) and that this anxiolysis is mediated by the conversion of testosterone to neurosteroids that interact with GABA(A) receptors. In Experiment 1, a 10-min exposure to opposite-sex conspecifics significantly reduced both male and female anxiety 20 min later (as indexed by increased open-arm time on an elevated plus-maze) compared to control mice not receiving this exposure. In contrast, locomotor activity (as indexed by enclosed-arm entries on the elevated plus-maze) was not significantly affected. The remaining experiments examined only male behavior. In Experiment 2, exposure to female urine alone was anxiolytic while locomotor activity was not significantly affected. Thus, urinary pheromones of female mice likely initiated the events leading to the male anxiolysis. In phase 1 of Experiment 3, sc injections of 500 microg of testosterone significantly reduced anxiety 30 min later while locomotor activity was not significantly affected. Thus, testosterone elevations were associated with reduced male anxiety and the time course consistent with a nongenomic, or very rapid genomic, mechanism of testosterone action. In phase 2 of Experiment 3, the anxiolytic effect of testosterone was dose dependent with a 250 microg sc injection required. Thus, testosterone levels likely must be well above baseline levels (i.e., in the range induced by pulsatile release) in order to induce anxiolysis. In Experiment 4, a high dosage of 5alpha-dihydrotestosterone was more anxiolytic than a high dosage of estradiol benzoate, suggesting that testosterone action may require 5alpha-reduction. In Experiments 5 and 6, 3alpha,5alpha-reduced neurosteroid metabolites of testosterone (androsterone and 3alpha-androstandione) were both anxiolytic at a lower dosage (100 microg/sc injection) than testosterone, supporting the notion that testosterone is converted into neurosteroid metabolites for anxiolytic activity. Experiments 7 and 8 found that either picrotoxin or bicucculine, noncompetitive and competitive antagonists of the GABA(A) receptor, respectively, blocked the anxiolytic effects of testosterone. However, conclusions from these 2 experiments must be tempered by the reduction in locomotor activity that was also seen. The possible brain locations of testosterone action as well as the possible adaptive significance of this anxiolytic response are discussed.