Abstract Background/Aims Traditionally viewed from the perspective of cartilage degeneration, osteoarthritis is increasingly seen as a disease of global joint dysfunction. Connective tissue extracellular matrix (ECM) is a crucial determinant of joint mechanobiology, providing cells with scaffolding, topographical cues, and a reservoir of soluble factors. While ECM dysregulation has been extensively studied in osteoarthritic cartilage, it remains poorly defined in other joint tissues. Here, we systematically review the composition, architecture, and remodelling of non-cartilage soft joint tissue ECM in human osteoarthritis and animal disease models. Methods A systematic search strategy was run through the MEDLINE, EMBASE and Scopus databases on 30 October 2020 and repeated on 1 October 2021. The search criteria included disease nomenclature, relevant tissues, as well as structural ECM components and architectural features. All papers were independently screened by two reviewers on the Covidence platform according to predefined eligibility criteria. Relevant clinical, demographic, and biological data were extracted from included studies, which were assessed for bias using the OHAT Risk of Bias Rating Tool for Human and Animal Studies. Results 148 of 8,156 identified studies met all eligibility criteria. 113 papers evaluated human osteoarthritis; of 35 animal studies, the most frequently used models involved surgical joint destabilisation in small mammals. ECM was best defined in menisci, ligaments, and synovium; fewer papers assessed skeletal muscles, tendons, and fat pads. Compared to the healthy joint, osteoarthritis is associated with qualitative and quantitative alterations in structural ECM components, most notably collagens and proteoglycans. In recent years, whole proteome sequencing has been employed to address these changes systematically. The mechanical properties of ECM change significantly in osteoarthritis in response to post-translational modifications, extensive calcification, and the marked loss of matrix organisation across the joint. Notably, some aspects of ECM remodelling in these tissues appear to precede discernible cartilage dysregulation. Similar ECM dysregulation is also observed in animal models, although inter-model variability in arthritogenic precipitant and the range of reported outcomes make comparisons difficult. Many studies are limited by significant bias, notably in the infrequent reporting of investigator blinding, and in the poor demographic matching of osteoarthritic and control patients. Encouragingly, the quality of methodology reporting and use of age-matched control populations have improved in recent years. Conclusion Current data provide compelling evidence of whole joint ECM changes in osteoarthritis and importantly suggest that these changes occur early in the disease process. How ECM dysfunction affects the behaviour of tissue-resident cells remains less well understood. Our work will support the design of disease-relevant biomaterials used to model osteoarthritis in vitro, helping to address this issue, by more accurately recreating the extracellular environment. Furthermore, the development of imaging modalities sensitive to connective tissue ECM changes warrants investigation from both diagnostic and prognostic perspectives. Disclosure J.Y. Mimpen: None. I.G.A. Raza: None. S.J.B. Snelling: None.