Fuel cell is an electro-chemical device that converts chemical energy into electrical energy. Optimal operation of solid oxide fuel cells require different components built around them. The physical structure of the fuel cell and associated components impose certain restrictions on the operation of the system. Thus control of a fuel cell system is performed by nonlinear model predictive control which is able to handle multivariate, objective and constraints. This work describes the application of nonlinear model predictive control applied on the fuel cell system by utilizing estimated states from the unscented Kalman filter. A brief description of the orthogonal collocation method as a means to discretize the nonlinear continuous model has also been provided in this work. (C) 2009 Elsevier Ltd. All rights reserved.