The evidence for a viral etiology of MS has been reviewed. The strongest evidence favoring a virus is based primarily on epidemiologic considerations. Less convincing evidence comes from pathologic studies, serology, lymphocyte reactivity to viral antigens, and reports of identification of virus in MS tissues. Animal models of viral demyelination exist, which may provide insight into possible etiologic agents and pathogenetic mechanisms. Considering all the data, it is clear that no agent can be convincingly linked to MS at the present time. If a single virus causes the majority of cases of MS, then a morbilliform virus--canine distemper--is a leading contender, because this agent is consistent with the epidemiologic and serologic findings and is highly neurovirulent for animals ranging from mice to primates. Since no virus fulfills the usual criteria for linking an infectious agent to a disease, other possibilities must be considered. If MS is caused by a single virus, it may be a common virus not presently considered as being associated with MS, or an agent as yet unidentified. It is also conceivable that multiple agents, acting alone or in concert, initiate the MS process, perhaps through a common immune-mediated pathway. In this regard, another human demyelinating disease--the Guillain-Barré syndrome--which may in some instances become a chronic remitting and relapsing disorder, is thought to be initiated by multiple infectious agents but to have an immunologic pathogenesis.