In a large population‐based multiple sclerosis (MS) sample (n = 7,951), 116 twins were identified (1.4%), of which it was possible to contact 97. In 27 pairs, the twin of the patient with MS was dead. Among the 70 remaining pairs, 16 were excluded because either the index patient or their twin did not accept protocol requirements. The final sample included 54 pairs. Twin zygosity was determined by DNA fingerprint analysis. We found 17 monozygotic pairs and 37 dizygotic pairs. One of the 17 monozygotic pairs (5.9%) and 1 of the 37 dizygotic pairs (2.7%) were concordant for MS. In 42 of the clinically unaffected twins, magnetic resonance imaging (MRI) of the brain was performed, and 9 scans were classified as indicative of MS. Three of these 42 patients had abnormal visual‐evoked potentials, but all 3 had normal MRI scans. The overall proportion of pairs in which the co‐twin had some form of clinical, radiological, or electrophysiological abnormality was approximately 30%; this proportion was independent of zygosity.