The occurrence of multiple sclerosis (MS) in twins has not previously been studied in complete nationwide data sets. The existence of almost complete MS and twin registries in Denmark ensures that essentially unbiased samples of MS cases among twins can be obtained. In this population-based study, virtually all Danish MS cases among twins born before 1983 with onset of MS after 1948 and diagnosis before I January 1997 were identified. Of 13 286 MS cases, 178 were twins and, of these 164 twin pairs were discordant and seven were concordant. We found significantly higher proband-wise concordance among monozygotic twins than dizygotic twins, with estimated proband-wise concordances of 24% (95% confidence interval (CI): 5-39%) for monozygotic and 3% (95% CI: 0-8%) for dizygotic twins. Thus, a monozygotic twin whose co-twin has MS has a 24% risk of developing the disease, while the corresponding risk for a dizygotic twin is only 3%. Our results largely confirm previously published concordance estimates and indicate that genetic factors are of importance in susceptibility to MS.