Summary The ultimate aim in the treatment of autoimmune disease is to restore self-tolerance to the autoantigen(s) in question. In lieu of this ideal result, the conversion of a destructive or pathogenic autoimmune response into one of benign autoimmunity would also be highly desirable. In either case the use of the antigenic epitope, which is the target of the destructive immune response, would ideally be employed so as to give speci®city to the protection without the need for long-term immunosuppression. This review describes a number of di erent approaches using various forms, doses, and routes of injection of speci®c neuroantigen to inhibit the di erent clinical varieties of autoimmune encephalomyelitis in a number of animal models; all done with the view to translating the ®ndings into the clinic for the treatment of multiple sclerosis. We conclude that any treatment strategy for multiple sclerosis (MS) must have a number of features: it must be clinically acceptable, speci®c, longlasting, require only short-term treatment, able to shut o ongoing disease, and have the potential to prevent or deal with epitope spreading. Few of the approaches we describe ful®ll all of these criteria. We suggest that investigations of new adjunctive agents to be used with a speci®c antigen be pursued, and that currently the use of chimeric proteins or DNA vaccination with or without the new adjunctives may hold the most hope for the future.