Background and purpose — It is unclear whether previous osteosynthesis is a risk factor for inferior outcome following shoulder arthroplasty for a proximal humeral fracture. We used data from the Danish Shoulder Arthroplasty Registry (DSR) to examine this question. Patients and methods — All 285 patients treated with a shoulder arthroplasty after failed osteosynthesis of a proximal humeral fracture reported to DSR from 2006 to 2013 were included. Each case was matched with 2 controls (570) treated with a primary shoulder arthroplasty for an acute proximal humeral fracture. Patient reported outcome was assessed using the Western Ontario Osteoarthritis of the Shoulder index (WOOS) and the relative risk of revision was reported. Results — The mean WOOS was 46 (SD 25) for a shoulder arthroplasty after failed osteosynthesis and 52 (27) after a primary shoulder arthroplasty. The relative risk of revision for a shoulder arthroplasty after failed osteosynthesis was 2 with a primary arthroplasty for fracture as reference. In a separate analysis of patients treated by locking plate the mean WOOS was 46 (24), with a relative risk of revision at 1.5 with a primary arthroplasty as reference. Interpretation — Compared with primary arthroplasty for proximal humeral fracture, we found an inferior patient-reported outcome and a substantial risk of revision for patients treated with a shoulder arthroplasty after failed osteosynthesis for a proximal humeral fracture. The risk and burdens of additional surgery should be accounted for when deciding on the primary surgical procedure.