The study of antibodies to the glutamate AMPA receptor subtype 3 in the pathogenesis of severe epilepsy associated with Rasmussen's encephalitis helped to crystallize the concept of autoimmune epilepsy. This work has been used as a paradigm to investigate autoimmunity, especially the humoral components, in several severe childhood epilepsies and more recently, in some of the more common acquired adult epilepsies. In addition, it helped prompt many open-label trials of immunotherapy for various epilepsies. We discuss the evidence that antiglutamate AMPA receptor subtype 3 antibodies are pathogenic and review the findings of the studies of immunotherapy for epilepsy. We conclude that immune treatments can have useful disease-modifying effects in some rare epilepsies, such as Rasmussen's encephalitis, at least in the short term and that their potential should be studied in the management of some forms of severe adult epilepsy.