Retrospectively, we assessed the specificity of two proposed magnetic resonance imaging (MRI) criteria for multiple sclerosis (MS) in patients suspected to have MS but who ultimately receive another diagnosis. Brain MRIs of 28 patients mixed with 28 MRIs of MS patients from the same cohort of 377 consecutively referred patients were scored by a neuroradiologist masked to the final diagnosis. The criteria for dissemination in space incorporated in the McDonald International Panel showed good specificity (89%). However, the more sensitive criteria proposed by a Subcommittee of the American Academy of Neurology resulted in a lower specificity (29%), indicating an increased risk of a false-positive diagnosis.