Strong arguments supporting a genetic linkage between susceptibility to HD and HLA are reported. These observations are based on data from 33 multiplex case families, gathered from international series and from our own studies. They confirm the disturbed segregation of HLA haplotypes among pairs of affected siblings (P less than 0.0005). An excess of a shared haplotype among first cousin pairs of patients is also observed (P less than 0.05). When both sib pairs and cousins are taken together, the segregation distortion is even greater (P less than 0.0002). Although the excess of HLA-identical affected sib pairs would favor a recessive mode of transmission of the disease, the lod score analyses do not allow one to conclude a simple genetic pattern. A two-gene model, based on epistatic cooperation, is discussed and could fit with an intermediate mode of transmission. The review of population data confirms the generally admitted trend of higher susceptibility borne by HLA-A1. There are converging arguments in favor of the prevalence of A1, B5, B18 in HD, and in particular A1 in the mixed cellularity form, while the haplotype, A1, B8, predominant in long survivors seems to possess a protective effect. It is expected that more data from multiplex families and prospective series of unrelated patients, fully HLA typed, may help to bring about a better understanding of the first reported HLA linkage with malignancy.