OBJECTIVES:  Post-acute care accounts for substantial amount of 90-day episode costs after joint replacement of the lower extremity. This study evaluated the impact of the first discharge destination on costs within 90-days of hospital discharge after primary total knee arthroplasty (TKA) or total hip arthroplasty (THA). METHODS: This retrospective cohort study included health care claims from the Centers for Medicare and Medicaid Services Standard Analytic Files, limited data sets (LDS). Patients aged 65 or older who had an inpatient TKA or THA between January 1, 2012 and October 1, 2014 were included. Patients’ discharge status (to home with or without home care versus skilled nursing facility) was the primary exposure variable and post-acute care costs was the outcome of interest. A multivariable linear regression model was used to estimate post-acute costs after adjusting for patient demographic and clinical characteristics, type of surgery (TKA vs THA), and the costs of index surgery. RESULTS: A total of 881,197 patients with total joint arthroplasty were identified. Of these, 37.1% were 75 years of age or older, 63.9% were women and 66.4% had TKA. The average Charlson comorbidity index of the population was 0.71. Most patients were in the South (36.7%) followed by Midwest (28.4%), West (18.5%) and Northeast (16.4%) regions. More than half of the patients (57.2%) were discharged home after their index surgery. Total post-acute costs were $9,400 lower ($3,975 vs $13,375, P<0.0001) for patients discharged home versus other settings of care. CONCLUSIONS: Costs for Medicare patients discharged directly to home (with or without home care) were 70% lower compared to those for patients discharged to other settings. Further work is required to determine the extent to which social, economic, and clinical factors allow for safe discharge home after joint replacement.