The epidemiology of multiple sclerosis suggests that a complex interaction of genes and environment contribute to susceptibility. To enrich for families with large genetic effects and to potentially reduce genetic heterogeneity, we screened a sample of 18,794 probands and identified forty families with four or more affected individuals. Within these 40 families, HLA DRB1*15 was present in 70% of affected individuals; the transmission disequilibrium test showed a significant excess in transmission of DRB1*15 alleles to affected individuals (47 transmitted, 19 untransmitted, chi (2) = 11.9, p = 0.00057). A 10 cM genome scan was performed and analyzed for linkage under a parametric model with heterogeneity. No excess of significant sharing was observed (HLOD > 3.3) in the parametric multipoint analysis. No region exceeded that for marker GATA8A05 with an HLOD = 1.11. Follow-up genotyping with 17 microsatellites revealed a significant two-point parametric HLOD = 3.99 at marker D4S1597. Transmission disequilibrium tests for markers in this candidate region showed no transmission distortion. A scan for variants in a gene adjacent to D4S1597, PALLD, was negative for synonymous or nonsynonymous changes. A final multipoint scan incorporating all microsatellites in the region provided an HLOD = 1.30. The inability to find significant linkage in these highly penetrant families suggests that linkage is not the optimal tool for dissecting the inheritance of MS.