PURPOSE OF THE STUDY Periprosthetic joint infections in total knee arthroplasty (TKA) represent one of the most limiting factors of implantation. Frequency of this complication is up to 2.5% in primary implantation. Revision TKA with the use of DAIR (Debridement, Antibiotics and Implant Retention) procedure is a widely accepted method in treating infection, but the indication criteria have not been clearly defined as yet. The lack of uniformity prevails also with respect to the surgical technique and the importance of respective techniques for successful treatment. The purpose of this study was to evaluate the factors affecting the twoyear survival of TKA after treating the infection by DAIR. MATERIAL AND METHODS We conducted a monocentric retrospective analysis involving 52 cases of infected TKA managed with DAIR in the period between 2007 and 2016. The evaluation took into account such factors as the sex, age, history of revision surgery for aseptic or septic reasons, and pathogens. The patients were divided into groups based on the McPherson criteria. As to the procedure, we monitored the effect of administered antibiotics, time interval between the manifestation of symptoms of TKA infection and surgery, exchange of modular parts, and use of pulse lavage, continual lavage, local antibiotic carrier, or combination of these techniques. Treatment failure was defined as persistent infection and transition to chronic suppressive antibiotic therapy or need for revision surgery of the respective joint due to recurrent infection of TKA, or death directly associated with the treatment of infected TKA in the follow-up period of 2 years after DAIR. The R software (Team Development Core, 2017) was used to carry out the statistical analysis. The target variable was the failure at two years after surgery. The Generalized Linear Model (GLM) was used for the binary dependent variable - the socalled logistic model with a logit link function. RESULTS 32 of 52 patients (61.5%) were successfully treated, of whom 18 women (62.1%) and 14 men (60.9%). The effect of causative agent, administered antibiotics, polyethylene insert exchange, McPherson score or history of revision surgery of the respective joint for aseptic reasons was not confirmed. The history of revision surgery for infection of the affected joint had a strong negative impact on treatment success, 10 of 13 (76.9%) implants failed as against 10 of 39 (25.6%) implants with negative history of infection. The mean time from surgery to the manifestation of infection was 5.9 weeks (0.5-47.5). When surgery was performed within 2 weeks from the manifestation of infection, 1 of 15 (6.7%) cases failed. In case of a later surgery, 19 of 37 (51.4%) cases failed. As concerns the used surgical technique, 60% (9/15) failure was reported in case of the combination of pulse lavage and continual lavage, 36.4% (4/11) in case of the combination of pulse lavage and local antibiotic carrier, 25% (4/16) in case of separate continual lavage, and 66.7% (2/3) in case of continual lavage with local antibiotic carrier. DISCUSSION The importance of individual factors in revision surgery of periprosthetic joint infections of TKA remains unclear. The world literature indicates as a major negative effect the time factor, the positive history of infection of the affected implant, or other previous revision surgery for aseptic reasons. Ambiguous results are achieved in assessing the effect of the pathogen, administered antibiotics or presence of fistula, the statistical significance of which has not been confirmed in our study. Questionable is also the importance of individual surgical techniques. CONCLUSIONS DAIR is a suitable method in treating infections of stable TKA without the history of revision surgery for infection. The surgery should be performed within 2 weeks from the manifestation of symptoms. Key words: debridement, antibiotics, infection, implant retention, total knee arthroplasty.