The coordinated temperature-pressure increase dynamic process during the start-up of the primary loop in a nuclear power plant is a core control challenge characterized by strong coupling, nonlinearity, and strict safety constraints, which are crucial for operational safety and economy. This study proposes and thoroughly explores a coordinated control strategy based on Model Predictive Control (MPC). This strategy achieves robust tracking of the target trajectories of temperature and pressure by constructing an accurate process dynamic model and online rolling optimization of multivariable control sequences. Compared with traditional manual control and conventional PID strategies, the proposed MPC scheme significantly improves the system’s dynamic response performance. Under strict guarantees of reactor core safety boundaries (such as avoiding coolant boiling and thermal stress exceeding limits), it effectively shortens the time for temperature and pressure increase, significantly enhancing the process stability and control precision. Additionally, MPC’s adaptive optimization capability reduces the intensity of operator intervention, enhancing the system’s robustness against condition fluctuations. The research findings provide effective theoretical basis and engineering practice references for the intelligent and autonomous operation control of key transient processes in nuclear power plants.