Abstract- Three-dimensional joint moments estimated during level walking can be useful in providing a mechanistic understanding of changes in gait patterns due to neuromuscular pathology. The accuracy of estimation of joint moments is influenced by several factors related to the measurement systems as well as the characteristics of the model used to describe the kinematics of lower extremities. In this paper using sensitivity analysis, we examine the effects of uncertainties in joint center estimation and the definition and construction of embedded axes on both the pattern and magnitude of hip, knee and ankle joint moments about the three anatomical axes. Kinematic and Kinetic data of representative normal subjects were used in the sensitivity analysis. First, hip, knee and ankle joint centers were perturbed 20, 10, and 10mm, respectively in the three coordinate directions and joint moments were recomputed for each case. Magnitude of peak joint moments varied from 7 to 33%. Next, the flexion-extension axis (embedded axis) at each joint was perturbed + or -15 degrees in the transverse plane from reference position and moments were recomputed. The magnitude of peak moments varied from 6 to 114% and the location of peak moment within the gait cycle also varied by as much as 36% of the gait cycle. In view of these results, it may be important to take into account the effect of uncertainties in the estimation of joint center and definition of embedded axes in interpreting joint moment data.