Aims. Currently, the US Center for Medicaid and Medicare Services (CMS) has been testing bundled payments for revision total joint arthroplasty (TJA) through the Bundled Payment for Care Improvement (BPCI) programme. Under the BPCI, bundled payments for revision TJAs are defined on the basis of diagnosis-related groups (DRGs). However, these DRG-based bundled payment models may not be adequate to account appropriately for the varying case-complexity seen in revision TJAs. Methods. The 2008-2014 Medicare 5% Standard Analytical Files (SAF5) were used to identify patients undergoing revision TJA under DRG codes 466, 467, or 468. Generalized linear regression models were built to assess the independent marginal cost-impact of patient, procedural, and geographic characteristics on 90-day costs. Results. A total of 9,263 patients (DRG-466 = 838, DRG-467 = 4,573, and DRG-468 = 3,842) undergoing revision TJA from 2008 to 2014 were included in the study. Undergoing revision for a dislocation (+$1,221), periprosthetic fracture (+$4,454), and prosthetic joint infection (+$5,268) were associated with higher 90-day costs. Among comorbidities, malnutrition (+$10,927), chronic liver disease (+$3,894), congestive heart failure (+$3,292), anaemia (+$3,149), and coagulopathy (+$2,997) had the highest marginal cost-increase. The five US states with the highest 90-day costs were Alaska (+$14,751), Maryland (+$13,343), New York (+$7,428), Nevada (+$6,775), and California (+$6,731). Conclusion. Under the proposed DRG-based bundled payment methodology, surgeons would be reimbursed the same amount of money for revision TJAs, regardless of the indication (periprosthetic fracture, prosthetic joint infection, mechanical loosening) and/or patient complexity. Cite this article: Bone Joint J 2020;102-B(7):959–964