Progabide (PGB), a gamma-amino-butyric acid receptor agonist, was administered, according to an open-label long-term design, to 40 adult patients suffering from complex partial seizures, with or without secondary generalization, whose response to carbamazepine (CBZ) monotherapy was unsatisfactory. A reference-baseline period of two months with carbamazepine monotherapy was followed by a two-month "add-on" period where increasing doses of progabide were added without modifying the CBZ regimen; then CBZ was withdrawn over 15-60 days and patients were followed up to 12 months' progabide treatment. Twenty-seven patients completed the trial but 12 of them had to be returned to CBZ + PGB bitherapy due to an increase of seizures following CBZ withdrawal. A definite therapeutic effect could be observed in nine patients on PGB monotherapy and in six patients on CBZ + PGB bitherapy. Side-effects of clinical relevance occurred in three cases and were represented by remarkable anxiety in two patients and a rise in serum glutamic oxalo-acetic acid and pyruvic transaminases with clinical symptoms of liver dysfunction in one, with rapid recovery following progabide discontinuation. In conclusion, progabide was effective against complex partial seizures in about 40% of patients not responding satisfactorily to available antiepileptic drugs. Although the withdrawal of previous antiepileptic drugs was not possible in all patients, progabide monotherapy was sometimes more effective than CBZ monotherapy, and several patients in whom bitherapy had to be restored benefited from the association of progabide.