Abstract: Coal catalytic gasification is a very complex process, which was affected by a lot of factors. A prediction model would be helpful for the understanding, design and optimization of such processing. Therefore, a prediction model for the weight loss, the initial temperature of gasification and the temperature of maximum gasification rate during coal catalytic gasification was established by using a three-layer improved Back Propagation (BP) neural network. The model prediction results indicate that the improved BP neural network has a high accuracy. Moreover, the maximum errors between the experimental and the predicted values are 5.18%, 5.65% and 2.33%, respectively, which are much smaller than those predicted by regression equation.