Background: Bicompartmental knee arthroplasty (BKA) was developed to treat medial tibiofemoral and patellofemoral osteoarthritis while preserving the anterior cruciate ligament to optimise knee kinematics. Our objective here was to compare the probability of achieving forgotten knee status and the functional outcomes at least two years after BKA versus total knee arthroplasty (TKA). We hypothesised that contemporary modular BKA produced better functional outcomes than TKA after at least two years, for patients with similar pre-operative osteoarthritic lesions. Material and methods: We conducted a two-centre prospective controlled study of 34 consecutive patients who underwent BKA between January 2008 and January 2011. Each patient was matched on age, gender, body mass index, preoperative range of knee flexion, centre, and surgeon to a patient treated with TKA. An independent observer evaluated all 68 patients after six and 12 months then once a year. Forgotten knee status was defined as a 1001100 value of the Forgotten Joint Score (FJS-12) and each of the five KOOS subscales. We also compared the two groups for knee range of motion, Knee Society Scores (KSSs), Timed Up-and-Go test (TUG), and UCLA Activity Score. Results: At a mean follow-up of 3.8 +/- 1.7 years, the probability of forgotten knee status was significantly higher in the BKA group (odds ratio, 4.64; 95% confidence interval, 1.63-13.21; 13=0.007, Chi(2) test). Mean post-operative extension was not significantly different between the groups, whereas mean range of knee flexion was significantly greater in the BKA group (130 degrees +/- 6 degrees vs. 125 degrees +/- 8 degrees after TKA; P=0.03). The BKA group had significantly higher mean values for the knee and function KSSs, TUG test, and UCLA score (P<0.04 for all four comparisons). Conclusion: After at least two years, contemporary unlinked BKA was associated with greater comfort during everyday activities (forgotten knee) and better functional outcomes, compared to TKA. These short-term results require validation in randomised trials with longer follow-ups. Level of evidence: III, case-control study. (C) 2015 Elsevier Masson SAS. All rights reserved.