Dissolved oxygen, a crucial water quality factor in aquaculture, requires precise control. This study proposes a Modified Smith fuzzy PID controller to tackle the control challenges posed by the time-varying, nonlinear, and time-delay characteristics of dissolved oxygen control systems. By incorporating a fuzzy PID algorithm into a Modified Smith predictor, this approach aims to mitigate the impact of model mismatch on dissolved oxygen control. Simulations were conducted under both matched and mismatched theoretical and actual model conditions. Results indicate that compared to conventional PID, fuzzy PID, and Smith fuzzy PID controllers, the Modified Smith fuzzy PID controller exhibits significantly lower maximum overshoot (0.78% on average) and achieves target dissolved oxygen concentrations in the shortest time (approximately 371 seconds). Experimental validation in real dissolved oxygen control processes further confirms its superiority. The Modified Smith fuzzy PID controller demonstrates the best control performance, with accuracy and robustness meeting the requirements for precise dissolved oxygen control in recirculating aquaculture systems.