The application of model predictive control (MPG) to complex, nonlinear processes results in a nonconvex optimization problem for computing the optimal control actions. This optimization problem can be addressed by discrete search techniques, such as the branch-and-bound method, which has been successfully applied to MPG. The discretization, however, introduces a tradeoff between the number of discrete actions (computation time) and the performance. This paper proposes a solution to these problems by using a fuzzy predictive filter to construct the discrete control alternatives. The filter is represented as an adaptive set of control actions multiplied by a gain factor. This keeps the number of necessary alternatives low and increases the performance. Herewith, the problems introduced by the discretization of the control actions are diminished. The proposed MPC method using fuzzy predictive filters is applied by the temperature control of an air-conditioned test room. Simulations and real-time results show the advantages of the proposed method.