The actions of the anticonvulsant gabapentin [1-(aminomethyl)cyclohexaneacetic acid, Neurontin] have been somewhat enigmatic until recently, when it was claimed to be a gamma-aminobutyric acid-B (GABA(B)) receptor agonist acting exclusively at a heterodimeric complex containing the GABA(B(1a)) splice variant (Mol Pharmacol 2001;59:144-152). In this study, we have investigated the effects of gabapentin on recombinant GABA(B(1a)) and GABA(B(1b)) receptors coexpressed with GABA(B(2)) in five different functional recombinant assays, its ability to inhibit [(3)H]GABA binding in a GABA(B) receptor-selective binding assay using rat synaptic membranes, and its ability to inhibit transient lower esophageal sphincter relaxations in Labrador retriever dogs. Up to a concentration of 1 mM, gabapentin displayed no agonistic effects on either the GABA(B(1a,2)) or the GABA(B(1b,2)) heterodimer, when these were expressed in Xenopus laevis oocytes or mammalian cells and assayed by means of electrophysiology, calcium mobilization, inositol phosphate, and fluorometry assays. Gabapentin did not displace [(3)H]GABA from GABA(B) receptor sites in rat synaptic membranes. Finally, in contrast to the classic GABA(B) receptor agonist baclofen, gabapentin was unable to inhibit transient lower esophageal sphincter relaxations in dogs. Because of high levels of GABA(B(1a)) in the canine nodose ganglion, this finding indirectly supports the inactivity of gabapentin on the GABA(B(1a,2)) heterodimer demonstrated in various in vitro assays. In light of these results, we find it highly questionable that gabapentin is a GABA(B) receptor agonist. Hence, the anticonvulsive effects of the compound have to arise from GABA(B) receptor-independent mechanisms. This also implies that the first GABA(B) receptor splice variant-selective ligand remains to be discovered.