Background and purpose - There is still no consensus on whether to use thromboprophylaxis as a standard treatment in shoulder replacement surgery. We investigated the use of thromboprophylaxis reported to the Norwegian Arthroplasty Register (NAR). The primary endpoint was early mortality after primary shoulder arthroplasty with and without thromboprophylaxis. Secondary endpoints included revisions within 1 year and intraoperative complications.Patients and methods - This observational study included 6,123 primary shoulder arthroplasties in 5,624 patients reported to the NAR from 2005 to 2018. Cox regression analyses including robust variance analysis were performed with adjustments for age, sex, ASA score, diagnosis, type of implant, fixation, duration of surgery, and year of primary surgery. An instrumental variable Cox regression was performed to estimate the causal effect of thromboprophylaxis.Results - Thromboprophylaxis was used in 4,089 out of 6,123 shoulder arthroplasties. 90-day mortality was similar between the thromboprophylaxis and no thromboprophylaxis groups (hazard ratio (HR) = 1.1, 95% CI 0.6-2.4). High age (> 75), high ASA class (≥ 3), and fracture diagnosis increased postoperative mortality. No statistically significant difference in the risk of revision within 1 year could be found (HR = 0.6, CI 0.3-1.2). The proportion of intraoperative bleeding was similar in the 2 groups (0.2%, 0.3%).Interpretation - We had no information on cause of death and relation to thromboembolic events. However, no association of reduced mortality with use of thromboprophylaxis was found. Based on our findings routine use of thromboprophylaxis in shoulder arthroplasty can be questioned.