Manipulations of GABA function have been found to produce highly variable effects in animal models of anxiety. In the present series, an ethological version of the murine elevated plus-maze was used to examine in detail the behavioural profiles of diazepam (1.5 mg/kg; positive control) and a range of GABA-related compounds: valproic acid (100–400 mg/kg), No-711 (1.25–10.0 mg/kg), muscimol (0.5–3.0 mg/ kg), (+)bicuculline (4.0–8.0 mg/kg), picrotoxin (0.25– 2.0 mg/kg), R(+)baclofen (0.375–3.0 mg/kg) and CGP 35348 (25–200 mg/kg). On both conventional and ethological indices, results confirmed the anxiolytic profile of diazepam under present test conditions, and revealed substantially similar effects for the GABA-T inhibitor, valproic acid (100–400 mg/kg), and the GABA_A receptor agonist, muscimol (2 mg/kg). The GABA reuptake inhibitor, No-711, produced weak anxiolytic-like effects at low doses (1.25–2.5 mg/kg) but disrupted behaviour at the highest dose tested (10 mg/ kg). Although the GABA_A receptor antagonists, (+)bicuculline and picrotoxin, produced changes indicative of anxiety enhancement, concomitant behavioural suppression was evident at high doses (8 mg/kg and 1–2 mg/kg, respectively). Further studies suggested that the effects observed with bicuculline may be mediated by an active metabolite, such as bicucine. In contrast to the effects of valproic acid and direct GABA_A receptor manipulations, the GABA_B receptor agonist, R(+) baclofen, non-specifically disrupted behaviour at the highest dose tested (3 mg/kg) while the GABA_B receptor antagonist, CGP 35348, was inactive over the dose range studied. Although present data confirm the sensitivity of the plus-maze to agents which modify GABA_A receptor function, further studies will be required in order more fully to characterize this relationship.