Epileptogenesis induced by electrical kindling of rats appears to be superior to the acute maximal electroshock seizure (MES) test in normal animals in predicting the efficacy and adverse effect potential of drugs in patients with partial epilepsy. Unfortunately, inclusion of such kindling models in primary screening is hampered by the laborious and expensive procedure of stimulation and recording with implanted brain electrodes. Furthermore the size of the rats excludes their use in initial testing where compound availability is often limited for the 'first batch synthesis'. The present study demonstrates that chronic electrical stimulation with corneal electrodes in mice can rapidly yield large numbers of kindled animals with a seizure phenomenology reflecting partial seizures in man. A pharmacological characterisation showed that corneally kindled mice can be used repeatedly for several drug experiments with reproducible results. The seizure protection and adverse effect potential obtained with proven antiepileptic drugs were similar to the effects observed in amygdala kindled rats and their corresponding clinical profile in partial epilepsy. Protection was obtained with vigabatrin and levetiracetam in this new model despite their lack of anticonvulsant activity in the acute MES test. Furthermore, in agreement with clinical findings with NMDA antagonists, MK-801 revealed more severe adverse effects in corneally kindled mice than in normal animals. These results suggest that corneal kindling of mice represents a sensitive and valid screening model for the identification of new therapies for partial epilepsy in man.