The systematic study of multiple sclerosis (MS) in populations, started in 1929 by Sydney Allison, now consists of over 400 publications dealing with the prevalence of MS throughout the world. However, any attempt to redefine the pattern of geographical differences in MS frequency remains as difficult as ever. The comparison of prevalence studies carried out in different areas and times is made difficult by the variability in surveyed population sizes, age structures, ethnic origins and composition, and the difficult quantification of numerators, especially regarding the recognition of benign and very early cases. Additionally, complete case ascertainment depends on access to medical care, local medical expertise, number of neurologists, accessibility and availability of new diagnostic procedures, the degree of public awareness about MS, and the investigators' zeal and resources. Critical examination of the more recent data on MS prevalence leads to some revisions of previously held concepts, the most interesting of which is the appreciation of the greater influence of genetic factors on disease acquisition. The rarity of MS among Samis, Turkmen, Uzbeks, Kazakhs, Kyrgyzis, native Siberians, North and South Amerindians, Chinese, Japanese, African blacks and New Zealand Maoris, as well as the high risk among Sardinians, Parsis and Palestinians, clearly indicate that the different susceptibilities of distinct racial and ethnic groups are an important determinant of the uneven geographic distribution of the disease. The updated distribution of MS in Europe, showing many exceptions to the previously described north-south gradient, requires more explanation than simply a prevalence-latitude relationship. Prevalence data imply that racial and ethnic differences are important in influencing the worldwide distribution of MS and that its geography must be interpreted in terms of the probable discontinuous distribution of genetic susceptibility alleles, which can however be modified by environment. Because the environmental and genetic determinants of geographic gradients are by no means mutually exclusive, the race versus place controversy is, to some extent, a useless and sterile debate.