In the standard inverse dynamic method, joint moments are assessed from ground reaction force data and position data, where segmental accelerations are calculated by numerical differentiation of position data after low-pass filtering. This method falls short in analyzing the impact phase, e.g. landing after a jump, by underestimating the contribution of the segmental accelerations to the joint moment assessment. This study tried to improve the inverse dynamics method for the assessment of knee moment by evaluating different cutoff frequencies in low-pass filtering of position data on the calculation of knee moment. Next to this, the effect of an inclusion of direct measurement of segmental acceleration using accelerometers to the inverse dynamics was evaluated. Evidence was obtained that during impact, the contribution of the ground reaction force to the sagittal knee moment was neutralized by the moments generated by very high segmental accelerations. Because the accelerometer-based method did not result in the expected improvement of the knee moment assessment during activities with high impacts, it is proposed to filter the ground reaction force with the same cutoff frequency as the calculated accelerations. When this precaution is not taken, the impact peaks in the moments can be considered as artifacts. On the basis of these findings, we recommend in the search to biomechanical explanations of chronic overuse injuries, like jumper's knee, not to consider the relation with impact peak force and impact peak moment.