This well-written paper presents a general predictive controller for nonlinear systems. The basic approach is to consider the nonlinear system as modelled by a Hammerstein model (a linear ARMA model cascaded with a nonlinear part that is approximated by a polynomial expressed in powers of the control input). The resulting control algorithm is a nonlinear form of generalized predictive control. The design procedure starts with the determination of a linear controller to provide a self-tuning generalized predictive controller for the ARMA part of the Hammerstein model. To determine the plant input control, the inverse of the nonlinear polynomial function has to be found. It is this inversion that causes Hammerstein-like approaches to be time-consuming and, at times, unreliable, since the solution of a nonlinear system (albeit a polynomial) has to be found for every recursion of the controller updating procedure. A novel procedure is presented here to simplify the root finding problem by taking only one step of a Newton-Raphson algorithm where the initial guess is given by the results of the previous iterations. This approach, also commonly used for other numerical analysis problems where a root solving step is an inner loop, has the advantage of requiring only a few calculations and of filtering out wild input variations. Since this is a heuristic approach, its usefulness has been ascertained by carrying out simulations and comparisons with nonlinear deadbeat control on the same plant moded. The approach proposed in this paper is claimed to give superior results.