In this thesis, the optimal maximum angle of pelvic yaw rotation was determined for a model of the humanoid robot DRC-Hubo+ developed by Hubolab, KAIST. In order to improve the energy usage and joint motor wear of the robot, in the form of minimising the squared joint velocity and the position variation respectively, and the work space of the robot. An equation for the pelvic yaw rotation was proposed and utilised in the analysis of the optimal angle. The walking path patterns used in the analysis are based on ZMP-based preview control and Orbital Energy. The robot kinematic model is constructed based on the denavit-hartenberg convention for forward and inverse kinematic. The dynamic robot model is determined numerically, and based on the Euler-Lagrange dynamic equation. The results obtained include a relation between the optimal angle and the step length, for the minimum squared velocity- and for the minimum position variation analyses.