Understanding the inheritance of rheumatoid arthritis (RA) has been the quest of intense investigation over the last decade. One major focus of these efforts has been the attempt to identify genes within the class I1 region of the major histocompatibility complex (MHC) that confer susceptibility for disease. These studies have been based largely on finding associations between class I1 serologic specificities and disease. The results have not been simple to interpret, and lucid analysis is made more difficult by the highly complicated nomenclature that is used to describe the class I1 HLA system. The advent of recombinant DNA technology, however, has greatly expanded our knowledge of this system over the last few years and has offered alternative interpretations of disease association data. One such interpretation, the shared epitope hypothesis, has been described previously (1) and is the subject of this review.To understand the shared epitope hypothesis, one must have some knowledge of the genetic organization of the class I1 region and of the biochemical structure of class I1 molecules. Briefly, the human class I1 region (Figure 1) extends to nearly l , OOO, OOO base pairs, and includes at least 14 different genes. With the exception of DOp and DZa, these genes are generally found in 1 of 3 major subregions: DR, DQ, or DP.Each subregion contains at least 1 functional a Figure 2. Alpha and beta chains from the DQ subregion are polymorphic (i.e., there are multiple alleles at each locus in the population), and together they encode the DQ serologic specificities (DQwl-3).The DR subregion also contains 2 functional p chain genes, designated DRPI and DRPIII. These p chains are both polymorphic. The D W I gene encodes the classic DR specificities (DR1-14); the DRPIII gene encodes the DRw52 and DRw53 specificities. The genes within the DR and DQ subregions are very closely linked and are almost always inherited together as a unit. Therefore, DR and DQ subregion alleles form stable haplotypes in the population. For example, the DR4 allele (encoded by the D W I gene) is almost always found in association with DRw53 (encoded by DRpIII) and DQw3 (encoded by DQ a and P), thus forming the typical DR4,DRw53,DQw3 haplotype. As discussed herein, all the variability between different DR4 haplotypes is located in the DRPI gene. The other linked genes, namely DRPIII, DQa, and DQP, are identical in all DR4 haplotypes, with few exceptions (2,3).The structural features of class I1 molecules at the cell surface are shown in Figure 2 (top). For comparison, class I molecules are also shown. In the case of class I1 molecules, both a and P chain molecules are inserted into the membrane and associate with each other in a noncovalent fashion to form an alp heterodimer. The class I1 p chain contains 2 immunoglobulin-like external domains. The first, or N-terminal, domain (Figure 2) is the site of most of thevariability