OBJECTIVE The purpose of this study was to assess the results of reimplantations of total knee arthroplasties complicated by infection. Outcome was assessed in terms of eradicated infection and function. METHODS This retrospective multicentric study included 107 cases of infected total knee arthroplasties treated by changing the implants. Seventy-seven patients had a two-stage revision and thirty had a one-stage procedure. Patients were reviewed with a minimal 2-year and an average 52-month follow-up. RESULTS Revision arthroplasty (one- or two-stage) eradicated infection in two out of three patients. With a two-year follow-up, revision arthroplasty was successful in 77% of patients without any sepsis risk factor, in 65% of patients with one risk factor and in 33% of patients with at least two risk factors. After reimplantation for total knee arthroplasty infection, overall function outcome was good (KS knee score: 74.8 after two-stage revision and 75.5 after one-stage revision, NS). After two-stage procedures, the knee outcome was excellent in one-third of patients, good in another third and fair or poor in the final third. After one-stage reimplantation, 40% of the knees had an excellent outcome, 30% a good outcome and 30% a fair or poor outcome. Regarding functional outcome, overall results were fair (KS function score 62.5 for one-stage and two-stage revisions). Functional outcome was fair or poor in 42% of patients with a two-stage procedure and in 55% of patients with a one-stage revision (NS). CONCLUSIONS Our study was unable to disclose any difference between one-stage and two-stage revision for eradicating infection. Unfavorable systemic and local conditions decreased the rate of success after revision total knee arthroplasty for infection. Length of infection before reimplantation, number of surgical procedures and bacterial virulence or resistance were not, in our series, predicting factors for failure of septic revision total knee arthroplasty. No difference was found for the clinical and functional results between one-stage and two-stage procedures. Functional outcome was fair or poor for half of the patients after septic revision total knee arthroplsty. The use of an external device between the two procedures for two-stage revision significantly decreased the functional outcome compared with the use of a spacer. Articulated spacers did not offered any advantage compared with a static spacer for functional outcome.