In this study, the use of an artificial neural network for predicting the reactivity of coal/char combustion was investigated. A database containing the combustion rate reactivity of 55 chars derived from 26 coals covering a wide range of rank and geographic origin was established to train and test the neural networks. The heat treatment temperature of the chars ranged from 1000 to 1500 degrees C and the combustion rate reactivity of the chars were measured using thermogravimetric analysis in a temperature range of 420-600 degrees C. Three correlation parameter sets were compared, which contained a coal rank parameter (either vitrinite reflectance or fixed carbon content), a parameter representing the extent of pyrolysis, combustion temperature, and char surface area. The results showed that when sufficient amount of training data are available, a neural network model can be developed to predict the combustion rates of coal chars with good accuracy and robustness. Fixed carbon content appeared to correlate better than random vitrinite reflectance Ro with combustion rates of coal chars. Total surface areas of the chars correlated to the combustion rates and when these values were used as one of the inputs to the neural network, better predictions were achieved.