To balance the intermittent renewable energy and cope with the increasing nuclear power installed capacity in China, it is necessary for nuclear power plants (NPPs) to operate with load-following mode for enhancing economic competitiveness. The modular high temperature gas-cooled reactor (MHTGR) is an appropriate candidate for load-following operation due to the advanced features such as robust fuel elements, online fueling and full-power-range temperature negative feedback. The MHTGR-based nuclear steam supply systems (NSSS) which produces superheated steam flow for electricity, is the core for any NPPs. The proper control of NSSS is the prerequisite for the safe, stable and efficient load-following operation. However, in current engineering practice, the set-point trajectories of neutron flux, primary coolant temperature, primary and secondary flowrates are just the curves by linking the referenced steady values from the thermal hydraulic design, which are further determined by the thermal power requirement. Thus, it is meaningful to investigate an optimization-oriented control method to improve the operation efficiency. Motivated by this, two dynamic matrix control (DMC) with cascade structure are presented to improve both thermal power and steam temperature of the MHTGR-based NSSS. The thermal power is optimized by the first DMC through adjusting the set-point of nuclear flux. The second DMC is then designed based on the integration of the NSSS, the proportional-integral-derivative (PID) and the first DMC to improve the steam temperature response by adjusting the feed-water flowrate. The control performance during wide range operation is shown and discussed with the simulation results of a large-scale MHTGR plant model. (C) 2019 Elsevier Ltd. All rights reserved.