In view of the shortcomings of the current prediction models of blasting fragmentation, artificial neural network theory was utilized to research on fragmentation prediction of ore stacking of in-situ blasting and leaching . According to the tectonic features of ore stacking by blasting for in-situ leaching, the three-layer feedforward neural network was established. Based on the practical engineers of home and abroad in-situ blasting and leaching, the artificial neural network model was trained by use of back propagation algorithm, so the nonlinear map relationship between fragmentation distribution of ore stacking by blasting and its influential factors was established. Furthermore, based the existing examples, the predicted results of the neural network were tested with ones of in-situ measurements, and the result has shown that the neural network model is feasible and exact for predicting the fragment size of ore stacking of in-situ blasting and leaching.