Background and purpose: There is controversy regarding the results of stemmed and stemless total shoulder arthroplasty (TSA) used for osteoarthritis. Therefore, we aimed to compare revision rates of stemmed and stemless TSA and to examine the impact of metal-backed glenoid components.Methods: We included all patients reported to the Danish Shoulder Arthroplasty Register from January 1, 2012 to December 31, 2022 with an anatomical TSA used for osteoarthritis. Primary outcome was revision (removal or exchange of components) for any reason.Results: 3,338 arthroplasties were included. The hazard ratio for revision of stemless TSA adjusted for age and sex was 1.83 (95% confidence interval [CI] 1.21–2.78) with stemmed TSA as reference. When excluding all arthroplasties with a metal-backed glenoid component, the adjusted hazard ratio for revision of stemless TSA was 1.37 (CI 0.85–2.20). For the Eclipse stemless TSA system, the adjusted hazard ratio for revision of a metal-backed glenoid component was 8.75 (CI 2.40–31.9) with stemless Eclipse with an all-polyethylene glenoid component as reference.Conclusion: We showed that the risk of revision of stemless TSAs was increased and that it was related to their combination with metal-backed glenoid components.