Objective To investigate the distribution characteristics of model numbers associated with the third-generation Oxford unicompartmental knee arthroplasty （UKA） prostheses， in order to provide evidence-based guidance for the selection of appropriate UKA prostheses in clinical practice. Methods A retrospective analysis was conducted on data from 270 patients who underwent UKA using third-generation Oxford prostheses and fulfilled the inclusion criteria between April 2019 and September 2023. Patient characteristics， including gender， age， height， weight， affected side， and prosthesis model numbers， were collected and recorded. The distribution and matching patterns of femoral and tibial prosthesis model numbers were evaluated. Multiple ordinal logistic regression analyses and receiver operating characteristic （ROC） curves were employed to identify significant predictors influencing the selection of femoral component models. Additionally， a parameter distribution map related to UKA prosthesis model numbers was developed. Results Among the femoral prostheses， the XS size accounted for 21.5%， S for 44.1%， M for 26.3%， and L for 8.1%. Regarding tibial prostheses， AA represented 11.1%， A 23.7%， B 29.3%， C 18.9%， D 10.4%， and E 6.7%. The matching rates between tibial and femoral components were as follows： 93.3% for tibial AA corresponding to femoral XS， 96.0% for tibial A/B corresponding to femoral S， 95.8% for tibial C/D corresponding to femoral M， and 91.7% for tibial E corresponding to femoral L， resulting in an overall matching rate of 95.2%. Height （AUC = 90.63%）， gender （AUC = 81.59%）， and weight （AUC = 84.47%） showed significant correlations with femoral model selection （P < 0.01）， whereas age and side did not demonstrate statistical significance. The S femoral model was predominantly used in female patients （59.6%）， while the M and L models were more commonly utilized in male patients （65.2%）. When female patients' height was ≤160 cm， the combined usage rate of the XS and S models exceeded 90%. For male patients taller than 175 cm， the utilization rate of the L model reached 88.9%. The accuracy rate of preoperative prediction based on height and gender was 70.7%， which improved to 77.0% when incorporating intraoperative tibial model information. Conclusions The distribution of UKA prostheses in the study population differed from that observed in Western populations， as well as from other regions in Asia and China. The distribution map of UKA prostheses and the corresponding matching rules established in this study demonstrate a high level of predictive accuracy， which can effectively support preoperative planning and intraoperative selection of femoral prosthesis model numbers.