We evaluated the usefulness of the MRI criteria for multiple sclerosis (MS) proposed by Paty et al and Fazekas et al in 36 Japanese MS patients, using HTLV-I-associated myelopathy (HAM)/tropical spastic paraparesis (TSP) as the control. Although 30 of 36 HAM/TSP patients had multiple white matter lesions on T2-weighted cranial MRI, only two fulfilled the MRI criteria for MS. At the same time, 31 of the 36 MS patients fulfilled the primary MRI criterion, yielding 93% specificity and 86% sensitivity for the criterion. MS has disease-specific MRI abnormalities.