Maintaining normal knee biomechanics is crucial for preserving joint health after anterior cruciate ligament reconstruction (ACLR), particularly in paediatric patients who face a higher risk of long-term complications from suboptimal surgeries. This study employed a neuromusculoskeletal-finite element (NMSK-FE) modelling pipeline to generate personalized boundary conditions for finite element (FE) models of three paediatric participants, each assessed eight months post-ACLR. For each participant, three FE model categories were developed: an intact, an actual post-surgery, and 135 simulated ACLR models representing systematic variations in surgical parameters (graft type, size, femoral tunnel placement, and pretension). Knee kinematics and tibial cartilage stresses were simulated during the stance phase of walking and compared with intact knee models using root-mean-square error (RMSE) and the coefficient of determination (R2). The ten best and ten worst surgical parameter sets were identified based on summed, normalized RMSE (nRMSE) across four kinematic and two cartilage-stress metrics. Applying the optimal surgical parameters from one participant to the others produced clear biomechanical deviations, with average RMSE increases of ∼28% in anteroposterior translation, ∼48% in internal/external rotation, and ∼24–28% in tibial cartilage stresses, underscoring strong patient-specific variability. Notably, actual surgical choices did not align with the top ten simulated scenarios, suggesting that conventional approaches may not yield optimal biomechanical outcomes. This study demonstrates the limitations of conventional techniques and advocates patient-specific NMSK-FE modelling to enhance surgical decision-making, with the potential to better restore knee biomechanics and reduce the risk of long-term complications.