The potency of eight standard anticonvulsants was tested in dose-response studies on kindled rats. Animals with either amygdala- or cortical-generalized seizures were used, and the effects of drugs were assessed on: (1) amygdala focal activity; (2) cortical focal activity; and (3) the generalized convulsion triggered from either focus. Ethosuximide, which is used against absence attacks, was not effective at subtoxic levels against any type of kindled seizure. The seven other drugs, all of which are effective against tonic-clonic seizures, were found to be: (a) most potent against generalized convulsive seizures; (b) slightly less potent against cortical focal activity; and (c) only partially effective against amygdala focal activity even at high (toxic) doses. The effect of these drugs on kindled seizures closely parallels their known clinical effects against (respectively) tonic-clonic, simple partial, and complex partial attacks. It is concluded that the kindling preparation could provide new pharmacological models for several different types of clinical seizure. Its most important use, however, is likely to be as a model for complex partial seizures, since there is at present no satisfactory pharmacological model for these common and drug-resistant attacks.