As neural networks are parallel machines, they hold great promise for solving complex problems. Due to the ability of neural networks to approximate arbitrary non-linearities, neural networks play a vital role in control field. To implement a neural network based predictive control, dynamic modeling of the process is required. The developed dynamic model is used to predict the output and try to minimize the difference between the predicted outputs and desired trajectory. A neural network based predictive control is implemented for a PMTD Process. © 2005 IEEE.