Distributed parameter systems (DPSs) constitute an important class of systems that include many industrial processes modeled by partial differential equations (PDEs). Model predictive control (MPC) techniques for DPSs have been relatively scarce because of the relative difficulty of solving PDEs. Conventional MPC approaches for DPSs have usually been developed on the basis of approximate lumped models. The resulting controllers can exhibit poor performance characteristics or require substantial on-line computation to ensure adequate performance. This paper presents a novel MPC scheme for output control of distributed parameter systems that is based on the method of characteristics. It is shown via simulation that the proposed approach can yield a high-performance controller with a comparatively small computational load.