Kainate-induced seizures and seizures induced by tossing stimulation in epilepsy-prone EL mice are considered as models of complex partial seizures. We used these models to evaluate the anticonvulsive effects of 2-[ N-(4-chlorophenyl)- N-methylamino]-4 H-pyrido[3.2-e]-1,3-thiazin-4-one (YM928), a novel alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor antagonist. In the kainate-induced seizure test in rats, wet-dog shakes (WDS) were reduced by oral administration of YM928 at doses of 7.5 mg/kg and 30 mg/kg. YM928 (15 mg/kg) reduced the number of WDS within the first 80 min, but then prolonged the time of occurrence compared with the other groups. Significant reduction in kainate-induced motor seizure was observed with 4-30 mg/kg. YM928 did not induce apparently abnormal behaviour at doses of 2-15 mg/kg but did induce sedation at 30 mg/kg. Carbamazepine (40 or 80 mg/kg), valproate (600 mg/kg), diazepam (2.5 mg/kg), and phenobarbital (20 or 40 mg/kg) exerted anticonvulsant effects against motor seizures, but only valproate, at a dose that also caused sedation, suppressed WDS. Phenytoin and ethosuximide did not show significant anti-kainate effects. In the tossing stimulation test in EL mice, i.p. injection of YM928 at 5 mg/kg or 10 mg/kg significantly increased the number of stimulations required to elicit generalized seizure. Carbamazepine (4 or 8 mg/kg), phenytoin (8 or 16 mg/kg), valproate (100-400 mg/kg), diazepam (0.5 mg/kg), phenobarbital (1.3 or 2.5 mg/kg) and ethosuximide (75-300 mg/kg) exerted significant anticonvulsant effects against these seizures. These results indicate that YM928 has anticonvulsant effects on seizure models that are characteristic of partial onset seizures in humans. YM928 is expected to have beneficial effects against human complex partial seizure with secondary generalization or temporal lobe epilepsy.