We have discussed partially mutually exclusive, partially overlapping models for the pathogenesis of the spondylarthropathies. Not all possibilities have been presented here; others are discussed elsewhere (77, 78). Furthermore, we have not addressed the issue of B27-negative spondylarthropathy. However, in our opinion, the key to understanding the pathogenesis of the spondylarthropathies lies in the interaction between the class I MHC molecule HLA-B27 and the T cell response. Although a T cell response driven by persisting bacterial antigen is still an attractive hypothesis, it does not explain all the known aspects of spondylarthropathy pathogenesis. The possibility of autoimmunity triggered by bacterial infection needs also to be considered, especially the new idea of HLA-B27-derived peptides presented by class II MHC molecules. The predominant involvement of joints is not easily explained in the case of autoimmunity. Cross-reactivity to joint-specific structures such as type II collagen (79) and/or bacteria inside the joint at the beginning of the immune response, with induction of local autoimmunity, might be involved. Most of the issues raised here could be tested by experiment, and we can expect to learn soon whether any of these models will explain the pathogenesis, or if we have to look further. The PCR technique will facilitate the search for bacteria not only in peripheral joints, but also now in sacroiliac biopsy samples from patients with AS and other spondylarthropathies. A prospective study on ReA in an endemic area should teach us more about predisposing factors (for example for Shigella-induced enteritis, which occurs in many parts of the world outside Europe and the US) (80).(ABSTRACT TRUNCATED AT 250 WORDS)