The proportion of patients over the age of 90 years continues to grow, and the anticipated demand for total joint arthroplasty (TJA) in this population is expected to rise concomitantly. As the country shifts to alternative reimbursement models, data regarding hospital expenses is needed for accurate risk-adjusted stratification. The aim of this study was to compare total in-hospital costs following primary TJA in octogenarians and nonagenarians, and to determine the primary drivers of cost. This was a retrospective analysis from a single institution in the U.S. We used time-drive activity-based costing (TDABC) to capture granular total hospital costs for each patient. 889 TJA’s were included in the study, with 841 octogenarians and 48 nonagenarians. Nonagenarians were more likely to undergo total hip arthroplasty (THA) (70.8% vs. 42.4%; p < 0.0001), had higher ASA classification (2.6 vs. 2.4; p = 0.049), and were more often privately insured (35.4% vs. 27.8%; p = 0.0001) as compared to octogenarians. Nonagenarians were more often discharged to skilled nursing facilities (56.2% vs. 37.5%; p = 0.0011), experienced longer operating room (OR) time (142 vs. 133; p = 0.0201) and length of stay (3.7 vs. 3.1; p = 0.0003), and had higher implant and total in-hospital costs (p < 0.0001 and 0.0001). Multivariate linear regression showed implant cost (0.700; p < 0.0001), length of stay (0.546; p < 0.0001), and OR time (0.288; p < 0.0001) to be the strongest associations with overall costs. Primary TJA for nonagenarians was more expensive than octogenarians. Targeting implant costs, length of stay, and OR time can reduce costs for nonagenarians in order to provide cost-effective value-based care.