The paper presents use of an artificial neural network (ANN) for predicting the thermal-flow behavior of a solid oxide fuel cell with no algorithmic solution merely by utilizing available experimental data. The error backpropagation algorithm was used for an ANN training procedure.