Maintaining desired concentration of the dissolved oxygen (DO) in an activated sludge process is crucial for feasible and efficient operation of a wastewater treatment plant. The dissolved oxygen dynamics is nonlinear and of high dimension. The available models involve many parameters that are very difficult to estimate. Utilising the dynamics structure and its multiple time scale a simplified nonlinear SISO model was recently adopted with a disturbance inputs that can be efficiently and sufficiently accurately predicted over short time period. Based on this model a nonlinear model predictive controller was designed showing good performance. However, necessity of solving a nonliner optimisation task during the controller operation limits its performance under large and fast changes of disturbances or reference trajectories. In the paper a fuzzy Takagi - Sugeno type model of the nonlinear dynamics is produced based on its local linearisations. The recently proposed fuzzy predictive control strategy is then applied to obtain a nonlinear fuzzy predictive controller. The controller is tested and validated on physical data sets showing substantial savings in computing time with negligible loss on its performance.