Sustained electrical stimulation of the perforant pathway was used to induce long-lasting hippocampal seizures in conscious rats. One hour prior to stimulation, rats were given i.p. injections of either saline or a commonly used antiepileptic drug, carbamazepine (5H-dibenz[b, f]azepine-5-carboxamide; CBZ; 20 mg/kg). When tested 2 weeks later in a water maze, both the saline- and the carbamazepine-pretreated rats showed similarly a severe impairment in spatial learning compared to non-stimulated controls. Histological evaluation revealed that the pyramidal cell damage was (P < 0.05) milder in the carbamazepine-pretreated group in the CA1, but not the CA3c subfield. However, the number of somatostatin-immunoreactive neurons in both stimulated groups was reduced equally. Thus, at the dose of 20 mg/kg, which is a usual anticonvulsive dose in humans, carbamazepine seems to offer only partial protection against pyramidal cell damage, but no protection against the hilar somatostatin-immunoreactive neuron loss or the spatial learning deficit after perforant pathway stimulation in rats. The result clearly differs from that obtained either with a GABA (gamma-aminobutyric acid)-enhancing drug and a novel antiepileptic, vigabatrin (4-amino-hex-5-enoic acid) or with a competitive NMDA (N-methyl-D-aspartate) receptor antagonist, CGP 39551 (DL-[E]-2-amino-4-methyl-5-phosphono-3-pentenoic acid carboxyethylester) in the same test situation.