Linearised model-reference systems are used for integrating the controls of a nonlinear lumped-parameter boiler model. The model equations are based on mass and heat-balance, heat-transfer and momentum relations previously derived for the assumed sections of an actual operating oil-fired boiler.The study investigates the control of the process, represented by a set of nonlinear equations, using the structure of a linear model for generating an optimal sequence of discrete inputs with reference to a quadratic cost function.Integrated control related to the linearised models provides superior performance compared with conventional control of the system operating with step disturbances using suitably defined performance criteria. The results demonstrate the suitability of a linear model for control of the particular nonlinear system over relatively wide load levels. A reduced-order linear system also appears to be adequate for providing an integrated control structure, and would be suitable for forming the basis of an online adaptive digital-control system.