Background In a retrospective multicenter study, we evaluated the efficiency and outcomes of the different therapeutic options for infection after reversed shoulder arthroplasty. Methods Thirty-two patients were reoperated on for infection after reversed shoulder arthroplasty between 1996 and 2011. The mean age was 71 (55-83) years. The involved implants were primary prostheses in 23 cases and revision prostheses in 9 cases. The average preoperative Constant score was 34 (11-69). Six of these patients needed 2 successive procedures. A total of 38 procedures were performed: débridement (13), 1-stage (5) or 2-stage revision (14), or implant removal (6). At last follow-up (mean, 36 months; range, 12-137 months), every patient had clinical, biologic, and radiographic evaluation. Results Infections were largely caused by coagulase-negative staphylococci (56%) and Propionibacterium acnes (59%). The complication rate was 26%. At last follow-up, 26 patients were free of infection (81%). The final Constant score was 46 (12-75). After débridement with implant retention, the mean Constant score was 51 (29-75), but the healing rate was only 54%. Implant revision (1 or 2 stage) led to better functional results than implant removal (46 vs. 25; P = .001), with similar healing rates (73% and 67%, respectively). Patients with low initial impairment (Constant score > 30) were not significantly improved by surgical treatment. Conclusion Débridement is the less aggressive option but exposes patients to healing failure. It should be proposed as a first treatment attempt. Revision of the implant is technically challenging but preserves shoulder function, with no higher rate of residual infection compared with implant removal.