Data are reviewed on the sib risk, the dizygotic twin concordance rate and the prevalence rate of multiple sclerosis (MS) in twins. Concordant affected twin pairs are so rare that it is unlikely that a single study (no matter how large) will establish whether they nevertheless are more common than expectation based on the null hypothesis of sib risk. For this reason, it is necessary to consider all the studies in which twin pairs were ascertained in series. There now seem enough data to cast doubt on this null hypothesis. Moreover, the prevalence of MS in twins seems low. Both these suggestions would be explained by the hypothesis that MS is a rare sequel of late exposure to common childhood infection.