Ultrasound-based ligament strain estimation shows promise in non-invasively assessing knee joint collateral ligament behavior and improving ligament balancing procedures. However, the impact of ultrasound-based strain estimation residual errors on in-silico arthroplasty predictions remains unexplored. We investigated the sensitivity of post-arthroplasty kinematic predictions to ultrasound-based strain estimation errors compared to clinical inaccuracies in implant positioning.Two cadaveric legs were submitted to active squatting, and specimen-specific rigid computer models were formulated. Mechanical properties of the ligament model were optimized to reproduce experimentally obtained tibiofemoral kinematics and loads with minimal error. Resulting remaining errors were comparable to the current state-of-the-art. Ultrasound-derived strain residual errors were then introduced by perturbing lateral collateral ligament (LCL) and medial collateral ligament (MCL) stiffness. Afterwards, the implant position was perturbed to match with the current clinical inaccuracies reported in the literature. Finally, the impact on simulated post-arthroplasty tibiofemoral kinematics was compared for both perturbation scenarios. Ultrasound-based errors minimally affected kinematic outcomes (mean differences < 0.73° in rotations, 0.1 mm in translations). Greatest differences occurred in external tibial rotations (−0.61° to 0.73° for MCL, −0.28° to 0.27° for LCL). Comparatively, changes in implant position had larger effects, with mean differences up to 1.95° in external tibial rotation and 0.7 mm in mediolateral translation. In conclusion, our study demonstrated that the ultrasound-based assessment of collateral ligament strains has the potential to enhance current computer-based pre-operative knee arthroplasty planning.