A total of 267 families with two or more siblings with multiple sclerosis (MS) were genotyped with 14 restriction fragment length polymorphisms at the TCR β locus. A nonparametric linkage analysis of the data showed no evidence for linkage to this locus (mlod=0.11). No significant allelic or haplotype transmissions were observed in the total sample of 565 patients. After stratification for the presence of HLA DRB1*15, an association was observed between the BV25S1*1-BV26S1*1-BV2S1*1 haplotype and MS (P=0.00089). This was not significant upon correction for multiple comparisons. It was also not significant when the haplotype frequency in affected individuals was compared to a normal control sample (P=0.77). Furthermore, the associated haplotype was followed-up in an independent sample of 97 nuclear families with a single DRB1*15-positive child with MS. The BV25S1*1-BV26S1*1-BV2S1*1 haplotype did not show significant evidence for transmission distortion but the same trend was seen (P=0.21). There were no significant associations observed in the DRB1*15-negative patients and no detectable difference was seen in the DRB1*15-positive BV25S1*1-BV26S1*1-BV2S1*1 association when comparing different subgroups based on clinical course of MS. These results show no evidence for linkage and fail to establish an association between MS susceptibility and the TCR β locus.