OBJECTIVE The classical twin study has the potential to evaluate the relative contribution of genes and environment and guide further research strategies, provided the sampling and methods of analysis are correct. We wish to review all the more informative twin studies on multiple sclerosis (MS). METHODS We examined six large population-based twin studies in MS and calculated indices of heritability (h(2)), which is the traditional method of assessing genetic contribution to disease and to allow comparison between studies. RESULTS This index was found to vary widely from 0.25 to 0.76 with large confidence intervals that reflect small sample size and prevent robust interpretation. CONCLUSION Overall the studies support a genetic contribution to disease; however, the imprecision of the heritability estimates and potential biases that they contain mean that very little inference can be drawn its exact size. Given that the magnitude of genetic effect cannot be measured because of the relative infrequency of MS; the consequent difficulty in collecting an informative sample; and in many countries, the lack of a comprehensive twin register, we suggest that further twin prevalence surveys should not be undertaken. Twin studies could be used more effectively in other ways, such as the co-twin case-control approach.