The hemochromatosis gene (HFE) maps to 6p21.3, in close linkage with the HLA Class I genes. Linkage disequilibrium (LD) studies were designed to narrow down the most likely candidate region for HFE, as an alternative to traditional linkage analysis. However, both the HLA-A and D6S105 subregions, which are situated 2-3 cM and approximately 3 Mb apart, have been suggested to contain HFE. The present report extends our previous study based upon the analysis of a large number of HFE and normal chromosomes from 66 families of Breton ancestry. In addition to the previously used RFLP markers spanning the 400-kb surrounding HLA-A, we examined three microsatellites: D6S510, HLA-F, and D6S105. Our combined data not only confirm a peak of LD at D6S105, but also reveal a complex pattern of LD over the i82 to D6S105 interval. Within our ethnically well-defined population of Brittany, the association of HFE with D6S105 is as great as that with HLA-A, while the internal markers display a lower LD. Fine haplotype analysis enabled us to identify two categories of haplotypes segregating with HFE. In contrast to the vast majority of normal haplotypes, 50% of HFE haplotypes are completely conserved over the HLA-A to D6S105 interval. These haplotypes could have been conserved through recombination suppression, selective forces and/or other evolutionary factors. This particular haplotypic configuration might account for the apparent inconsistencies between genetic linkage and LD data, and additionally greatly complicates positional cloning of HFE through disequilibrium mapping.