Most detailed knowledge of the MHC outside of mammals has come from studies of chickens, originally due to the economic importance of the poultry industry. We have used our discoveries about the chicken MHC to develop a framework for understanding the evolution of the MHC, based on the importance of genomic organisation for gene co-evolution. In humans, MHC class I molecules are polymorphic and determine the specificity of peptide presentation, while the molecules involved in antigen processing are functionally monomorphic. The genes for tapasin, transporters associated with antigen presentation (TAPs) and inducible proteasome components (LMPs) are located in and beyond the class II region, far away from the class I genes in the class I region. In contrast, chickens express only one class I locus at high levels, which can result in strong MHC associations with resistance to particular infectious pathogens. The chicken TAP and tapasin genes are located very close to the class I genes, and have high levels of allelic polymorphism and moderate sequence diversity, co-evolving their specificities to work optimally with the dominantly expressed class I molecule. The salient features of the chicken MHC are found in many if not most non-mammalian species examined, and are likely to represent the ancestral organisation of the MHC. Comparison with the MHC organisation of humans and typical mammals suggests that a large inversion brought the class III region into the middle of the MHC, separating the antigen processing genes from the class I gene, breaking the co-evolutionary relationships and allowing a multigene family of well-expressed class I genes. Such co-evolution in the primordial MHC was likely responsible for the appearance of the antigen presentation pathways and receptor–ligand interactions at the birth of the adaptive immune system. Of course, much further work is required to understand this evolutionary framework in more detail.