Abstract: The oxygen excess ratio (OER) of the proton exchange membrane fuel cell (PEMFC) reflects the balance between the air supply flow and the actual hydrogen oxygen chemical reaction. To address the OER control problem, we propose a proportional-integral-derivative (PID) control algorithim whose parameters are turned by a secondary performance index radial basis function (RBF) neural network. To adjust the PID controller parameters, as in the established PEMFC air system model, we use the sum of the system output OER error and the control voltage increment as performance indexes of the neural network coordination algorithm. A comparison of the simulation results of the proposed secondary performance index with those of the single OER error index RBF-PID controller and the traditional PID controller shows that the OER overshoot is reduced by approximately 10% compared with that of the single OER error index RBF-PID controller and by approximately 30% compared with that of the PID controller. The control voltage also changes more quickly and reasonably.