Context To assess the effect of a surgical procedure on a patient, it is conventional to use clinical scores before and after the procedure, but it is increasingly common and recommended to weight the results of these scores with the notion of minimal clinically important difference (“MCID”). This MCID should be determined using either the data distribution method based on score variation, or the anchor method, which uses an external question to categorize the results. MCIDs vary from one population to another, and to our knowledge there has been no investigation in France for total knee arthroplasties (TKAs). We therefore conducted a prospective study on a population of TKAs in order to: 1) Define MCID in France on a population of TKAs for the Oxford score, KOOS (Knee injury and Osteoarthritis Outcome Score) and its derivatives, 2) Determine whether MCID for these scores in France is comparable to results in the literature. Hypothesis Is the MCID for total knee arthroplasty in France comparable to other results in the literature? Material and method This was a prospective observational study in which 218 patients (85 men, 133 women) with a mean age of 72 years [27–90] who had undergone a primary TKA out of 300 initially included responded, before and after surgery, to the Oxford-12, KOOS and Forgotten Joint Score (FJS) questions (mean follow-up 24 months). MCID was calculated using the distribution method as well as the anchor method (“improvement 1 to 5” and “improvement yes or no”). Results At a mean follow-up of 24 months [18–36], the Oxford-12 score increased from 16 ± 8 [0–41] to 34 ± 11 [6–48] (p < 0.001), all components of the KOOS score were improved and the FJS at follow-up was 47 ± 32 [0–100]. For the anchor “improvement 1 to 5”, there were 14 unimproved patients, 23 patients in identical condition and 179 patients improved by surgery. For the anchor “are you improved yes/no”, there were 8 unimproved patients, 22 in identical condition and 187 surgically-improved patients. The mean MCID for all methods (anchor method and distribution) was 10 [7–13] for Oxford-12, 12 [12–12] for KOOS Symptom, 14 [12–17] for KOOS Pain, 12 [11–14] for KOOS Function, 14 [12–16] for KOOS Sport, 15 [15–16] for KOOS Quality of Life (QOL), 11 [10–12] for KOOS 12, 15 [12–18] for KOOS 12 Pa in. 12 [12–13] for KOOS 12 Function, 15 [15–15] for KOOS 12 QOL, 14 [13–14] for KOOS Physical Function Short-form (PS) and 14 [13–16] for KOOS Joint Replacement (JR). Discussion The MCID for the Oxford-12, KOOS and its derivatives scores in a French population is comparable to that observed in other populations in the literature. Level of evidence IV; prospective study without control group.