The diagnosis of multiple sclerosis is based on dissemination in time and space. Before 2010 lack of evidence for dissemination in space could be substituted by a paraclinical test, cerebrospinal fluid (CSF) oligoclonal bands (OCBs). The present meta-analysis (13,467 patients) shows that the diagnostic specificity of OCB drops from 94% to 61% if inflammatory etiologies are considered. Importantly, this was not caused by poor laboratory practice. This review on CSF OCB further illustrates the conceptional problem of substituting dissemination in space with a biomarker. The potential prognostic value of intrathecal OCB will need to be tested prospectively.