The cytokines IL-23 and IL-17 have an important role in the pathogenesis of, and as a therapeutic target in, both animal models of chronic inflammation and some human chronic inflammatory diseases. The traditional view is that a main source of IL-17 is T cells and that IL-17 production is under the control of IL-23. IL-17 inhibition has shown good efficacy in clinical trials for ankylosing spondylitis (AS), a subtype of axial spondyloarthritis (axSpA) characterized by radiographic evidence of sacroiliitis. On the basis of data from animal models, genetic studies and the investigation of tissue and blood samples from patients with AS, IL-23 had also been predicted to be important in the pathogenesis of this disease and was therefore considered a potential therapeutic target for axSpA. However, two placebo-controlled, double-blind clinical trials in axSpA of monoclonal antibodies directed against either the p40 protein or the p19 protein of the IL-23 molecule had clear negative results. These findings indicate that IL-23 and IL-17 are at least partly uncoupled in axSpA. Reasons as to why, when and how such an uncoupling might occur are discussed in this Review, with special reference to the unique microenvironment of the subchondral bone marrow in axSpA.