Artificial neural network (ANN) approaches for modeling of proton exchange membrane (PEM) fuel cells have been investigated in this study. This type of data-driven approach is capable of inferring functional relationships among process variables (i.e., cell voltage, current density, feed concentration, airflow rate, etc.) in fuel cell systems. In our simulations, ANN models have shown to be accurate for modeling of fuel cell systems. Specifically, different approaches for ANN, including back-propagation feed-forward networks, and radial basis function networks, were considered. The back-propagation approach with the momentum term gave the best results. A study on the effect of Pt loading on the performance of a PEM fuel cell was conducted, and the simulated results show good agreement with the experimental data. Using the ANN model, an optimization model for determining optimal operating points of a PEM fuel cell has been developed. Results show the ability of the optimizer to capture the optimal operating point. The overall goal is to improve fuel cell system performance through numerical simulations and minimize the trial and error associated with laboratory experiments.