Background: The treatment of complex proximal humerus fractures in elderly patients is not yet fully elucidated. Of all treatment options, reverse shoulder arthroplasty (RSA) and non-operative treatment (NOT) appear to provide the best results. Evidence to guide the choice between the two is sparse. Therefore, this review provides an overview of the available evidence on RSA versus NOT. Methods: Studies comparing complex proximal humerus fractures in patients aged >65 years treated either with RSA or NOT were included for systematic review and direct comparison via pooled analysis of patient-rated outcome and range of motion. Indirect comparison of case series and non-comparative studies on either treatment was performed separately. Results: Three comparative studies including 77 patients treated with RSA and 81 treated non-operatively were analysed. The RSA group scored better for both the Constant–Murley score (mean difference 6 points) and DASH score (mean difference 8 points). No differences were detected in ASES, PENN score, pain scores, or range of motion between treatment groups. The most common complications for RSA were infection (3%), nerve injury (2%), and dislocation (2%). Reoperation was required in 5%. In the NOT group, common complications included malunion (42%), osteonecrosis (25%), and non-union (3%); no reoperation was required. Patient satisfaction was equal in both groups. Conclusions: The functional outcomes and range of motion after RSA seemed satisfactory and potentially superior to NOT in elderly patients. Patient satisfaction was comparable despite a high malunion and osteonecrosis rate in the non-operative treatment group, which did not require re-interventions.