Classical screening tests (maximal electroshock, MES, and threshold pentylenetetrazol, PTZ) employ non-epileptic rodents and identify antiepileptic drugs (AEDs) with mechanisms of action associated with significant CNS side effects. Thus MES identifies drugs acting on Na+ channels that produce cerebellar toxicity. It may be possible to produce novel AEDs more selectively targeted at voltage-sensitive (VS) ion channels. There is little specific evidence for the likely success of this strategy with subunit selective agents targeted at the different VS Na+ channels. Drugs targeted at specific VS Ca++ channels (T, N, P/Q types) may be useful in generalised seizures. There are many as yet unexplored possibilities relating to K+ channels. GABA related drugs acting on PTZ clonic seizures tend to induce sedation and muscle hypotonia. Studies in mice, particularly with knock-in mutations, but also with subunit selective agents acting via the GABA(A) benzodiazepine site, suggest that it is possible to produce agents which do or do not induce particular side effects (sedative, hypnotic, anxiolytic, muscle relaxant, amnesia, anaesthesia). Whether these findings transfer to man has yet to be established. Acquired epilepsy in rodents (e.g. kindling or spontaneous seizures following chemically- or electrically-induced status epilepticus) or acquired epilepsy in man (following prolonged febrile seizures or traumatic brain injury) is associated with multiple changes in the function and subunit composition of ion channels and receptor molecules. Optimal screening of novel AEDs, both for efficacy and side effects, requires models with receptor and ion channel changes similar to those in the target human syndrome.