This paper presents a scheme for designing a robust decentralized controller for an industrial utility boiler system. First, a new method for designing linear robust decentralized controllers for a class of nonlinear systems is presented. For systems in this class the feedthrough matrix, D is not equal to zero. By using a descriptor system representation, the system is converted into a system with D = 0. Then, sufficient conditions for closed-loop stability, and asymptotic output tracking over the operating-range of the system are obtained. Satisfaction of these conditions is formulated as an optimization problem with LMI constraints. By solving the optimization problem, a decentralized controller for the system can be calculated. In order to control a utility boiler, an identified model of the utility boiler is considered. The identified model is nonlinear, second order, and captures all the characteristics of the real system, including shrink and swell phenomenon, non-minimum phase behavior, and instability. Using the proposed methodology, a decentralized robust controller is designed for the identified model. Thereafter, the designed controller is applied to the simulation model of the real system. The simulation results show excellent input tracking and disturbance rejection. (C) 2008 Elsevier Ltd. All rights reserved.