During the last few years, the concept of multiple sclerosis (MS) as a pure inflammatory disease mediated by myelin reactive T cells has been challenged. Neither the specificity nor the mechanisms triggering or perpetuating the immune response are understood. Genetic studies have so far not identified therapeutic targets outside the HLA complex, but epidemiological and immunological studies have suggested putative pathogenetic factors which may be important in therapy or prevention, including the Epstein-Barr virus and vitamin D. Advances in the treatment of MS have been reached by manipulating the immune response where the pathogenesis of MS intersects experimental autoimmune encephalomyelitis, most recently by blocking T-cell migration through the blood-brain barrier. Antigen-specific approaches are effective in experimental models driven by a focused immune response against defined autoantigens, but MS may not fit into this concept. Novel candidate autoantigens which are not constitutively expressed in the brain, such as protein alpha-B crystallin or IgG V-region idiotopes, as well as evidence of pathogenetic heterogeneity and complexity, suggest that treating MS by tolerizing the immune system against an universal MS antigen may be a fata morgana. Further characterization of MS subtypes may lead to individualized treatment. However, shared immunological features, such as intrathecal production of oligoclonal IgG, suggest that potential therapeutic targets may be shared by most MS patients.