The purpose of this study was to identify factors associated with variation in direct costs with shoulder arthroplasty. This was a retrospective study of all shoulder arthroplasties performed at a single facility between July 1, 2011, and November 30, 2016. We collected patient factors, indications, procedure (including implant details), implant brand (A, B, and other), and complications. We collected direct costs over a 90-day period using a validated internal tool. We identified patient and procedure characteristics associated with costs using multivariable generalized linear models. A total of 361 patients were included, 19% with revision arthroplasty procedures, 32% with anatomic total shoulder arthroplasties, and 66% with reverse total shoulder arthroplasties (RTSAs). Of total costs, 13% were operative facility utilization costs and 58% were operative supply costs. Factors associated with increased total cost included younger age (P = .002) and an indication for surgery of other, that is, not osteoarthritis, a failed arthroplasty, or the sequelae of a rotator cuff tear (P = .030). Factors associated with increased operative costs included younger age (P = .002), use of an RTSA (P < .001), use of a bone graft (P < .001), implant brand B (P = .098), implant brands other than A and B (P = .04), the sequelae of a rotator cuff tear as an indication for surgery (P = .041), or an indication for surgery of other (P = .007). Most short-term (90-day) costs with shoulder arthroplasty are operative costs. Nonmodified factors associated with increased cost included younger age and less common indications for surgery, whereas potentially modifiable factors included the intraoperative use of a bone graft, implant brand, and RTSA use.