Résumé : Purpose: This finite element study aims to compare the biomechanical performance of three total knee arthroplasty (TKA) alignment strategies—mechanical alignment, kinematic alignment and inverse kinematic alignment—in knees with different degrees of varus deformity. We hypothesise that alignment choice has limited impact in rather neutral alignment but significant biomechanical effects in mild deformity. Methods: Two osteoarthritic patients scheduled for TKA, one with almost neutral alignment (1.5°) and one with mild (10.4°) varus deformity, were selected. For each patient, mechanical alignment, kinematic alignment and inverse kinematic alignment strategies were virtually applied using a posterior-stabilised fixed-bearing implant. Patient-specific knee joint geometries, including bones and ligaments, were modelled from CT scans. Gait simulations were conducted following ISO standard. Outputs analysed and compared among alignments included contact areas and contact forces, von Mises stress on the tibial insert, tibial anterior-posterior translation and tibial internal-external rotation. Results: In the almost neutral knee, all three alignments showed comparable kinematics and kinetics. However, in the mildly deformed case, notable differences emerged. Kinematic alignment achieved more balanced medial-lateral contact areas and forces, while inverse kinematic alignment displayed increased internal-external rotation, indicating potential instability from suboptimal ligament balance. Mechanical alignment resulted in higher medial stress, while kinematic alignment redistributed stress more evenly. Inverse kinematic alignment exhibited intermediate behaviour, resembling mechanical alignment in force distribution under mild deformity. Conclusion: Alignment strategy plays a critical role in TKA biomechanics, especially in case of patients with mild varus deformity. While similar outcomes are achievable with almost no deformities, patient-specific morphology heavily influences performance under more complex geometries. These findings support the adoption of a personalised alignment approach tailored to individual anatomical characteristics rather than rigid adherence to a single alignment philosophy. Level of Evidence: N/A.