40 multiplex multiple sclerosis (MS) families were analyzed for evidence of an MS susceptibility gene linked to the HLA region of the sixth chromosome. We assumed that population associations between specific HLA alleles and MS are due to linkage disequilibrium, so preference was given to hypotheses compatible with tight linkage, and explanations were sought for conflicting evidence. The analyses proceeded in two steps: (1) segregation analysis using the computer program POINTER, and (2) linkage analysis using LINKAS, first assuming linkage equilibrium and then allowing for linkage disequilibrium and etiological heterogeneity. The results of the segregation analyses were indeterminate. The results of the linkage analyses suggest that analyses that do not allow for disequilibrium lose substantial evidence on linkage. Of the models that were investigated under linkage disequilibrium, the best fit is with a model of complete linkage (theta= 0.0) in 75% of the pedigrees and no linkage in the remaining pedigrees. We were unable, however, to statistically reject another model involving loose linkage and no heterogeneity.