This paper describes application of nonlinear long range predictive control to an open loop unstable continuous stirred-tank reactor. For the controller, various nonlinear autoregressive integrated moving average with external input (NARIMAX) models were used and the model parameters were calculated by applying Bierman algorithm. For this calculation, a pseudo random binary sequence (PRBS) was given to the heat removal from the system in the short time domain before instability starts, reaction heat was chosen as a disturbance. The simulation program having mass and energy balances for reactor inlet and cooling system was used to obtain the model parameters and to design the tuning parameters of the nonlinear long range predictive control system. Heat removal from the system was chosen as a manipulated variable. The nonlinear long range predictive control algorithm provides a substantial array of tuning knobs, viz: minimum costing horizon (N-1), maximum costing horizon (N-2), control costing horizon (N-u), control weighting (lambda). The values of N-2 and lambda are varied in order to obtain the best result. The success of the various control action is estimated using the integral square of the error (ISE) and the integral of absolute value of error (IAE) criterions. It was observed that the value of N-2 with the best lambda chosen has the greatest effect on the control performance.