Inclusion body myositis (IBM) is often viewed as an enigmatic disease with uncertain pathogenic mechanisms and confusion around diagnosis, classification and prospects for treatment. Its clinical features (finger flexor and quadriceps weakness) and pathological features (invasion of myofibres by cytotoxic T cells) are unique among muscle diseases. Although IBM T cell autoimmunity has long been recognized, enormous attention has been focused for decades on several biomarkers of myofibre protein aggregates, which are present in <1% of myofibres in patients with IBM. This focus has given rise, together with the relative treatment refractoriness of IBM, to a competing view that IBM is not an autoimmune disease. Findings from the past decade that implicate autoimmunity in IBM include the identification of a circulating autoantibody (anti-cN1A); the absence of any statistically significant genetic risk factor other than the common autoimmune disease 8.1 MHC haplotype in whole-genome sequencing studies; the presence of a marked cytotoxic T cell signature in gene expression studies; and the identification in muscle and blood of large populations of clonal highly differentiated cytotoxic CD8+ T cells that are resistant to many immunotherapies. Mounting evidence that IBM is an autoimmune T cell-mediated disease provides hope that future therapies directed towards depleting these cells could be effective.