Digital human modeling and posture prediction can only be used as a design tool if the predicted postures are realistic. To date, the most realistic postures have been realized by simultaneously optimizing human performance measures (HPMs). These HPMs currently consist of joint discomfort, delta potential energy, and visual displacement. However these HPMs only consider the kinematics of human posture. Dynamic aspects of human posture such as external loads and mass of limbs have not yet been considered in conjunction with the current HPMs. This paper gives the formulation for a new human performance measure combination including the use of joint torque to account for the dynamics of human posture. Postures are then predicted using multi-objective optimization (MOO) techniques to optimize the combination of the new HPM and the current. The predicted postures are then compared with the benchmark postures which are those obtained from using the current HPMs only. Also examined is the effect the new HPM has on stability using Zero Moment Point (ZMP) stability criterion and the distribution of the ground reaction forces on the two feet.