The association of HLA-B27 with ankylosing spondylitis, a chronic inflammatory disease of the axial skeleton, and other spondyloarthropathies, is among the strongest of an MHC antigen and any disease. Yet, the basis for this association remains unknown. In this review the main current hypotheses concerning the pathogenetic role of HLA-B27 will be discussed. They focus on three molecular properties of the molecule: (1) its peptide-presenting specificity, (2) its slow folding and tendency to misfold, and (3) its capacity to form covalent heavy chain homodimers amenable to recognition by leukocyte receptors. On the basis of the peptide specificity spondyloarthropathies would be triggered through T-cell autoimmunity against a self-ligand of HLA-B27 elicited by a cross-reactive foreign antigen. HLA-B27 misfolding would trigger disease through activation of inflammatory pathways following induction of endoplasmic reticulum stress, thus independently of antigen presentation. Recognition of heavy chain homodimers by leukocyte receptors might be involved in disease through immunomodulation of both innate and adaptive responses to arthritogenic pathogens. None of these hypotheses can yet satisfactorily account for the pathogenesis of spondyloarthritides. It is proposed that the pathogenetic role of HLA-B27 will eventually be explained through a global understanding of its biology, in which the various features of this molecule are envisaged as inter-dependent in their contribution to disease.