Aims. To compare complication-related reoperation rates following primary arthroplasty for proximal humerus fractures (PHFs) versus secondary arthroplasty for failed open reduction and internal fixation (ORIF). Patients and Methods. We identified patients aged 50 years and over, who sustained a PHF between 2004 and 2015, from linkable datasets. We used intervention codes to identify patients treated with initial ORIF or arthroplasty, and those treated with ORIF who returned for revision arthroplasty within two years. We used multilevel logistic regression to compare reoperations between groups. Results. We identified 1624 patients who underwent initial arthroplasty for PHF, and 98 patients who underwent secondary arthroplasty following failed ORIF. In total, 72 patients (4.4%) in the primary arthroplasty group had a reoperation within two years following arthroplasty, compared with 19 patients (19.4%) in the revision arthroplasty group. This difference was significantly different (p < 0.001) after covariable adjustment. Conclusion. The number of reoperations following arthroplasty for failed ORIF of PHF is significantly higher compared with primary arthroplasty. This suggests that primary arthroplasty may be a better choice for patients whose prognostic factors suggest a high reoperation rate following ORIF. Prospective clinical studies are required to confirm these findings. Cite this article: Bone Joint J 2019;101-B:1272–1279