This paper presents and defines a new design optimization method for kinematic redundant robot manipulators based on their applications. Kinematic redundant manipulators can reach a pose with an infinite number of postures. So, identifying the best robot design and configuration for a set of desired tasks is a highly complex non-linear problem. This approach employs a task priority control algorithm to perform a task oriented robot design optimization. The design parameters are replaced by controllable prismatic or revolute virtual joints and controlled by the algorithm to accomplish the desired tasks. Therefore, this new method finds an optimal robot design for a set of tasks taking advantage of the robot kinematic redundancy. This method is evaluated on a highly kinematic redundant manipulator, which tracks a set of paths with its end-effector while maintaining good kinetostatic performance.