Operating temperature and humidity are key factors affecting the performance and lifetime of proton exchange membrane fuel cell (PEMFC) stacks. Aiming at the shortcomings of low control accuracy and poor robustness in existing water and thermal management strategies, a water and thermal management strategy based on the twin-delayed deep deterministic policy gradient (TD3) is proposed. The strategy was based on a water and thermal management system model, which improved the traditional control framework of independent control of the water pump, the radiator, and the cathode humidifier, and achieved synchronous control of temperature and humidity. Taking the water and thermal management of a PEMFC stack with 375 cells as an example, the proposed strategy is compared with the water and thermal management strategy based on the deep deterministic policy gradient (DDPG) through simulation experiments under the mixed driving cycle condition. The maximum deviations of the inlet temperature of the stack and the temperature of the stack from the ideal temperature under the TD3-based strategy are reduced by 0.10 K and 0.07 K, respectively, compared with the maximum deviations under the DDPG-based strategy; the TD3-based strategy maintains the cathode inlet humidity between 25% and 62%, while the DDPG-based strategy maintains the cathode inlet humidity between 0 and 98%. The results show that, compared with the DDPG-based strategy, the proposed strategy performs better, effectively reducing the maximum temperature deviation and the range of humidity changes of the stack.