Recent experiments have demonstrated that formation of functional type B gamma-aminobutyric acid (GABA(B)) receptors requires co-expression of two receptor subunits, GABA(B1) and GABA(B2). Despite the identification of these subunits and a number of associated splice variants, there has been little convincing evidence of pharmacological diversity between GABA(B) receptors comprising different subunit combinations. However, Ng et al. [Mol. Pharmacol., 59 (2000) 144] have recently suggested a novel and important pharmacological difference between GABA(B) receptor heterodimers expressing the GABA(B1a) and GABA(B1b) receptor subunits. This study suggested that the antiepileptic GABA analogue gabapentin (Neurontin) is an agonist at GABA(B) receptors expressing the GABA(B1a) but not the GABA(B1b) receptor subunit. The importance of this finding with respect to identifying novel GABA(B) receptor subunit specific agonists prompted us to repeat these experiments in our own [35S]-GTPgammaS binding and second messenger assay systems. Here we report that gabapentin was completely inactive at recombinant GABA(B) heterodimers expressing either GABA(B1a) or GABA(B1b) receptor subunits in combination with GABA(B2) receptor subunits. In addition, in both CA1 and CA3 pyramidal neurones from rodent hippocampal slices we were unable to demonstrate any agonist-like effects of gabapentin at either pre- or post-synaptic GABA(B) receptors. In contrast, gabapentin activated a GABA(A) receptor mediated chloride conductance. Our data suggest that gabapentin is not a GABA(B)-receptor agonist let alone a GABA(B) receptor subunit selective agonist.