Introduction <p>The frequency of revisions in total knee arthroplasty (TKA) is rising. Various classifications of bone defects exist, each with its own limitations. Recently, Belt et al. have proposed a new classification for TKA revisions based on X-ray imaging. We evaluated the Belt et al. classification and verified if this new classification is reliable, and if it correlates with the implant used during revision surgery for periprosthetic joint infection.</p> Methods <p>This is a retrospective study. We reproduced the paper proposed by Belt et al. with the radiological data of all patients who underwent two stage revision for infected TKA in our institution between January 2017 and December 2022. Five different operators classified the bone defect for each patient at two time points. Subsequently, we assessed intra- and inter-operator reproducibility. We also collect the surgery data from our registry to verify if there is a correlation between augment use and epiphyseal bone defect.</p> Results <p>The classification proposed by Belt is reliable, and have a good reproducibility inter and intraoperator. There is no correlation between the bone defect. And the use of augment, and so this classification is usless in the prediction of the material needed in the operating room.</p> Conclusion <p>The Belt at al. classification is reliable, but a classification which can predict the implant neded have to be developed.</p>