Collective evidence supports the notion that multiple sclerosis is principally an autoimmune disease. Much of it stems from models of experimental autoimmune encephalomyelitis, generated by inoculation of animals with central nervous system antigens such as MBP, PLP, S100 and MOG or peptides thereof. Different ways of immunization and different animal species and strains mirror different aspects of the neuro-pathology of multiple sclerosis, such as inflammation, demyelination or axonal damage, and reflect different clinical courses. In all these models, the first immune reactions take place in lymph nodes from which immune cells migrate into the circulation and then to the central nervous system. Adoptive transfer of myelin-reactive T cells from these animals produces pathology and disease in the central nervous system of naive healthy recipients. In the human disease, autoreactive T and B cells specific for a variety of central antigens are present in the immune repertoire. These cells appear to be activated in the periphery through a number of mechanisms which causes them to home to the central nervous system. Contact with the local immune circuitry of the brain stimulates clonal expansion of autoreactive T cells, initiating a cascade of immuno-inflammatory events in situ. Numerous ways of disrupting this complex sequence of events, either by non-specific immunosuppression or by targeting specific checkpoints, abrogate or ameliorate disease in animal models. All approved disease-modifying drugs have an impact on components of the systemic immune compartment. All have been shown to reduce the number of gadolinium-enhancing T1 lesions observed with magnetic resonance imaging, an index of acute inflammatory invasion of the central nervous system.