Diverse marker sets and validation techniques have previously been utilized, posing challenges in comparing studies when assessing soft tissue artefacts in knee joint kinematics from motion analysis. This study aimed to analyse the data obtained from three different marker sets with the results derived from radiostereometric analysis (RSA) in measuring angular movements of the knee joint. Twelve post-knee replacement participants performed a one-leg step-down movement. Knee joint angular movements were analysed in flexion–extension, adduction-abduction, and internal-external rotation across all marker sets. The results were subsequently compared with those obtained from the RSA system using simple linear regression, a linear mixed-effects model, mean values and mean differences. All marker sets were found to systematically underestimate flexion–extension compared to RSA, with differences intensifying at higher knee flexion angles. The mean differences in the sagittal plane between RSA and the IOR marker set, progressively increased from approximately 5° (95% CI 4.3–4.9) to 15° (95% CI 11.6–17.9), reaching a maximum difference of 20° (95% CI 13.8–25.7) at 40° of knee flexion. Transverse and frontal plane data from all marker sets exhibited erratic errors compared to RSA. In summary, knee flexion–extension motions were consistent between marker sets, indicating minimal impact on results based on the marker set choice. However, all marker sets systematically underestimated skeletal motions in knee flexion–extension compared to RSA measurements. Data from the transverse and frontal planes were too inconsistent and therefore not reliable for use.