BACKGROUND GABAA receptors containing an alpha2 subunit are proposed to mediate the anxiolytic effect of benzodiazepines (BZ) based on studies in transgenic mice using unconditioned models of anxiety. Conditioned models of anxiety were not assessed and are rarely encountered in phenotyping of genetically modified animals. The novel benzodiazepine site ligand L838,417 is a partial agonist at GABAA receptors containing an alpha2, alpha3 or alpha5 subunit and an antagonist at alpha1 receptors, giving an anxiolytic profile devoid of sedation. However, this compound has not previously been assessed in mice. OBJECTIVE (1) Establish the Vogel conflict test (VCT) in C57BL/6J mice and validate it with a range of pharmacological tools and (2) compare the full and partial GABAA receptor positive modulators chlordiazepoxide (CDP) and bretazenil (BRZ), respectively, with the subtype selective ligands zolpidem (ZOL; alpha1 selective) and L838,417. RESULTS (1) enhanced thirst (water deprivation or isoproterenol administration), analgesia (lamotrigine) or cognitive impairment (MK-801) did not generate false positives in the VCT; (2) CDP and BRZ engendered linear dose-related anti-conflict effects and also increased unpunished drinking; (3) L838,417 engendered a bell-shaped anti-conflict effect and did not increase unpunished drinking; (4) the anti-conflict effect of CDP and L838,417 were antagonised by flumazenil, whereas BRZ's effect was insensitive to this antagonist; and (5) ZOL induced motoric deficits and no anti-conflict effect. CONCLUSIONS We have established the VCT in C57BL/6J mice and validated this test behaviourally, physiologically and pharmacologically. The novel GABAA receptor ligand L838,417 was anxiolytic in this mouse model, and unlike the non-selective compounds, had no effect on unpunished drinking.