A haptic interface is a robotic force feedback device that provides a sense of touch to users of virtual reality simulations. This thesis presents a general method for the design optimization of parallel planar haptic devices based on maximizing the manipulability of the interface over its workspace. Manipulability is selected as the key design objective to ensure avoidance of singular configurations within the workspace and to maximize the interface's ability to generate feedback forces and torques in each direction in each handle location and orientation. The optimization approach developed in this thesis results in a set of candidate designs that are found by stepping the design parameters through the range of possible values, and testing the manipulability and other measures (including workspace area and space) at each location and orientation of the interface handle. To find the optimal design, a multi-objective approach is taken to generate a set of Pareto optimal designs. A smart Pareto filter is employed to yield a smaller set of designs representative of the full Pareto frontier. The most desirable design is chosen from this reduced set. The result is a general optimization method applicable to parallel haptic interfaces. The method is demonstrated on the design of a 3-RPR parallel planar interface.