During the operation of a nuclear power plant (NPP), the discharged water flows to the downstream purification unit after passing through the regenerative heat exchanger and the non-regenerative heat exchanger. To prevent the ion exchange resin in the purification unit from failing due to high temperature, it is essential to effectively control the discharge flow temperature. Due to the large time lag in the heat exchange process of the non-regenerative heat exchanger and the lag in temperature measurement instruments, the lead-lag PID control system currently used in engineering has unsatisfactory control performance, with periodic fluctuations in the steady-state temperature, accompanied by temperature approaching the alarm value. In this paper, the derivation of the formula for the primary-loop discharge flow temperature system is carried out based on the energy and mass conservation equations. Furthermore, a mechanism model of the primary-loop discharge flow temperature is constructed in MATLAB/Simulink. On this basis, a temperature control scheme based on model predictive control (MPC) is designed and constructed. The scheme is linearized at the operating point of temperature control in order to obtain the state space model of the discharge flow temperature, while the constraints of temperature and component cooling water (CCW) valve action are considered. Simulation tests are conducted on typical operating conditions and the control effect of MPC is compared and analyzed with lead-lag PID control. The simulation results demonstrate that MPC is superior to lead-lag PID control in terms of rapidity and anti-interference ability, greatly improving the problem of unsatisfactory control performance caused by large time lag.