Knee arthrodesis is used to treat patients with failed infected total knee arthroplasty (TKA). Among fixation methods, intramedullary nailing increases the chances of bone union but may carry a risk of infection around the nail. This risk is not well understood, because available case-series studies were not confined to patients with knee infection. Infection recurrence rates after knee arthrodesis with intramedullary nailing used to treat failed infected TKA are similar to those seen with other fixation methods. We retrospectively reviewed 31 cases of knee arthrodesis with fixation by a modular intramedullary nail performed at a subspecialized center treating complex osteoarticular infections (CRIOAC). The antibiotic regimen was determined based on multidisciplinary discussions and microbiological studies of preoperative and intraoperative specimens. Mean follow-up was 50 ± 22 months (range, 28-90 months). Arthrodesis was performed in one stage (n=6) or two stages (n=25). Success was defined as presence, after a postoperative follow-up of at least 24 months, based on the following criteria: normal erythrocyte sedimentation rate and/or C-reactive protein, no wound inflammation or sinus tract, no revision surgery, and no antibiotic treatment. Bone union was not a criterion for a successful arthrodesis procedure. Removal of the fixation material was required in three patients and long-term palliative antibiotic therapy in three patients (fixation material in place with repeated positive specimens) for a total of six failures due to infection (6/31, 19.4%). None of the patients experienced mechanical failure (no breakage of the material and no fixation failure of the nails designed to allow osteointegration). The mean leg length discrepancy was 10 ± 10 mm (range, 5-34 mm) and the mean Oxford score was 41 ± 11 (range, 23-58). The 50-month rate of arthrodesis survival to revision surgery for nail removal was 77.8 ± 4% and the 50-month rate of arthrodesis survival without revision surgery for persistent infection was 74.6 ± 4.2%. The infection recurrence rate was higher than with other fixation methods but remained acceptable (19.4%). Use of a modular intramedullary nail prevented major leg-length discrepancies, which are often poorly accepted by the patients, and allowed immediate weight bearing despite the often severe bone loss. Level IV, retrospective cohort study.