The single most important feature that determines the reliability of epidemiologic studies of multiple sclerosis (MS) is the use of well-defined, generally recognized diagnostic criteria. The long-accepted direct relationship between prevalence rates (PR) and latitude is no longer valid and has been replaced by the realization that genetic factors play an important role in the acquisition of the disease. The nature of environmental factors, which may be of more importance in influencing the clinical manifestations of the disease, remains obscure. Most potential risk factors that have been studied lack biologic plausibility. Biostatistical interpretation of epidemiologic data can only indicate possible causal relationships but cannot conclusively rule out their existence. Differences in PR among different ethnic groups in similar locations suggest that, in addition to genetic factors, there may be enhancing and/or protective influences which are probably environmental in nature. Migration studies have been interpreted as suggesting that MS is acquired before puberty and that age at migration is important in determining the risk of having clinical disease. Epidemics by age of onset vanish when recalculated on the basis of age of acquisition, and probably mean that more early and mild cases are being diagnosed. A new definition of PR is proposed, ie, onset-adjusted PR, which includes patients who had symptoms at the time of the survey, but had not yet been diagnosed as MS. However, individuals who had symptoms at the time of migration to a study area must not be included in PR.(ABSTRACT TRUNCATED AT 250 WORDS)