Reactor temperature control is very important as it affects chemical process operations and the product quality. Although PID controller, which is the linear controller and widely used in the chemical process industries, is able to control the temperature, the operating range is limited. Furthermore, its control performance when plant/model mismatches exist is not guaranteed. Recently, various advanced control techniques have been successfully applied to highly nonlinear systems. These include the Generic Model Control (GMC) and the Inverse-Model Control (IMC) techniques. However these methods still require reasonable and accurate process model and parameters, which are difficult to guarantee in many cases. For this reason we have used neural networks in conjunction with these methods to overcome this problem for the control of the reactor in this study. The neural network is used as a function estimator in the GMC method and as a model and controller in the IMC-PI method. Various simulations involving set point tracking and disturbance rejection under nominal and model-mismatch cases were performed using these hybrid methods. The results of these hybrid controllers were found to be better than the conventional PID and GMC methods in most cases. These results justify the use of the neural networks in such hybrid strategies as well as show their versatility in incorporating into the nonlinear control methods to cater for model mismatches and difficult to control process systems.