The remarkable association between HLA-B27 and ankylosing spondylitis (AS) remains an enigma. While previous reviews have discussed the controversies surrounding the involvement of bacteria in the etiology of this disease and the sequence variability between subtypes of HLA-B27, concepts of disease mechanism remain ill-defined. In this article Richard Benjamin and Peter Parham synthesize new data on the structure and function of HLA class I molecules into possible mechanisms that might underly the pathogenesis of AS.