A population-based study of multiple sclerosis (MS) was conducted in 2 northern Colorado counties in 1982 to determine MS prevalence, to compare the rates with recent North American surveys and to compare the methods used in these studies. Provisional cases were identified from: the patient rolls of MS service organizations, chart reviews in 2 neurology practices, a survey of physicians and a review of hospital discharge diagnoses. Crude-point prevalence for the 2-county region was 84 per 100,000. The age-adjusted rate was higher than the rate for the region above the 37th parallel projected from data in a 1976 national survey, but was comparable to rates obtained in localized surveys conducted in the northern tier of the country. The methodological results revealed that the highest yield sources were the MS service organizations and the neurology practice chart reviews. MS prevalence surveys which neglect these methods may underestimate MS prevalence by as much as 20-40%.