PURPOSE: Approximately 50% of knee osteoarthritis patients are eligible for unicompartmental knee arthroplasty (UKA), but only 10%-15% of knee replacements are UKAs. Higher UKA revision rates may prevent broader implementation. The hypothesis was that up to 40% of UKA revisions are potentially avoidable based on a radiographic analysis of indications, surgical technique and reasons for revision. METHODS: Consecutive UKA revisions between 2007 and 2022 from one high-volume UKA centre were analysed. Two independent reviewers systematically evaluated all preoperative, direct post-operative and prerevision radiographs (anteroposterior and lateral, stress radiographs if available). Using the Oxford group criteria, adequate use of UKA indications was assessed, as well as surgical technique errors and the presence of radiographic reasons for revision. Infections were excluded. RESULTS: Ninety-eight revisions were included with a median time to revision of 2.2 years (interquartile range: 0.9-5.5). UKA indications were satisfied in 45%. Presence of medial bone-on-bone osteoarthritis was doubtful or not present in 37%. Other indications were possibly not satisfied in 18%. Post-operative, major technical errors were identified in 7% of cases. No radiographic reason for revision was identified in 34%. Common reasons for revision were progression of lateral/patellofemoral osteoarthritis in 41%, bearing dislocation in 13% and periprosthetic fractures in 9% of cases. Uncemented fixation was associated with revision ≤2 years ( CONCLUSIONS: In conclusion, a substantial number (>
40%) of UKA revisions are potentially avoidable based on the present radiographic analysis. LEVEL OF EVIDENCE: Level III.