The recent discovery of autoimmune antibodies to the neuronal cell surface membrane and extra- or intra-cellular proteins, such as NMDAR and LGI1, shed light on a proposed new etiology of epilepsy, namely, "autoimmune epilepsy". A large part of this entity most likely belongs to a forme fruste of autoimmune (limbic) encephalitis. Seizures are usually subacute in onset and refractory to antiepileptic medications. Patients occasionally manifest multiple seizure semiologies, such as autonomic or faciobrachial dystonic seizures. They may develop other symptoms of autoimmune encephalitis to variable degrees. Clinical diagnosis of autoimmune epilepsy should be considered even before the results of antibody testing, when objective evidence of CNS inflammation, such as abnormal MRI (amygdala/hippocampal enlargement), FDG-PET (hypermetabolism), EEG (bitemporal spikes or subclinical seizure patterns) and/or CSF findings, are present. Early intensive immunotherapy could cure seizures and prevent the full-blown clinical manifestation of encephalitis. Some patients present with relatively mild clinical symptoms and/or MRI findings, and can have a smoldering course lasting years. Establishment of the biomarkers of the ongoing CNS inflammation, especially for the smoldering and the suspected autoantibody-negative cases, is warranted for tailored immunotherapy for both the subacute and chronic phases of the disease.