We have previously described a series of competitive GABA_A antagonists derived from the low-efficacy partial agonist 5-(4-piperidyl)-3- isoxazolol (4-PIOL, 4). The 2-naphthylmethyl analogue, 4-(2-naphthylmethyl)-5- (4-piperidyl)-3-isoxazolol (5), provided affinity for the GABA_A receptor site higher than that of the standard GABA_A receptor antagonist, SR 95531 (3). Molecular modeling studies of these compounds exposed a cavity at the receptor recognition site capable of accommodating aromatic groups of substantial size in the 4-position in the 3-isoxazolol ring. Here we present a series of analogues of 5, with various substituents in different positions in the naphthyl ring system (6a-k), and compounds with aromatic substituents directly attached to the 4-position of the 3-isoxazolol ring (71-n). The compounds have been pharmacologically characterized using receptor-binding assays and electrophysiological whole-cell patch-clamp techniques. All of the tested compounds show affinity for the GABA_A receptor site. While the 5-, 7-, and 8-bromo analogues, 6b-d, showed receptor affinities (K_i = 45, 109, and 80 nM, respectively) comparable with that of 5 (K_i = 49 nM), the 1-bromo analogue, 6a, provided the highest receptor affinity of the series (K_i = 10 nM). Introduction of a series of different substituents in the 1-position in the 2-naphthyl ring system led to compounds, 6c-k, with retained high affinity for the GABA _A receptor (K_i = 16-250 nM). Introduction of a phenyl ring directly into the 4-position on the 3-isoxazolol ring gave a 41-fold increase in affinity relative to that of 4-PIOL. In whole-cell patch-clamp recordings from cultured cerebral cortical neurons, all of the tested compounds were able to inhibit the effect of the specific GABA_A agonist isoguvacine, 6a showing antagonist potency (IC₅₀ = 42 nM) markedly higher than that of 3 (IC₅₀ = 240 nM). Molecular modeling studies, based on the compounds described, emphasized the importance of the distal ring in 5 for receptor affinity and the considerable dimensions of the proposed receptor cavity. Furthermore, the phenyl rings in 71 and in 6k were shown to represent highly favorable positions for an aromatic ring in previously unexplored receptor regions in terms of a pharmacophore model.