Nuclear magnetic resonance imaging (MRI) proved to be, from the first, a very sensitive method, allowing the visualisation of multiple sclerosis lesions, yet which never permitted to establish a non equivocal relationship between the semeiology of such lesions and the clinical signs. The multifocal aspect of disseminated multiple sclerosis lesions is probably one of several factors accounting for this discrepancy. The study of an autoimmune disease, experimental allergic encephalomyelitis (EAE), regarded as a suitable model for multiple sclerosis in humans, has been performed using MRI in order to unravel the pathogenesis of the disease and apprehend the mechanisms responsible for the formation of multiple sclerosis lesions. The study focused on the part played by the blood-brain barrier (BBB) in the induction process of an autoimmune disease, since the central nervous system is normally screened from immunological supervision, by this barrier. Models both of acute EAE, induced by active or passive transfer of the antigen (myelin basic protein-MBP)--and chronic EAE, induced by passive transfer of MBP-specific T cells and myelin glycoproteins or MOG-specific monoclonal antibodies, have been reproduced, and their evolution followed up using high field MRI. Every time, the crucial role of the BBB was evidenced by the synchronism existing between the clinical signs, the appearance of lesions, preferentially in the most sensitive or permeable areas, and the BBB breakdown encouraged by the action of adjuvants. The physiopathological study of EAE using MRI is suggestive of the concept of systemic disease for multiple sclerosis, according to a two-step process, involving, in a first stage some primary viral or bacterial infection, causing T-cells to be sensitized to the host's own proteins by molecular mimicry, and in a second stage some bacterial infection or accidental circumstances which, resulting in a BBB breakdown, would provide immunocompetent cells with an opportunity to reach their target.