Abstract: Reservoir evaluation are seriously short of physical parameters during development of the small Mutou oilfield due to some historical factors.Physical parameters with high accuracy are needed for analyzing residual oil distri-bution,constructng reservoir geologic models and numeri-cal modeling along with the advancing of development.A method is presented in this paper for predicting reservoir porosity and permeability with artificial neural network BP model.Compared with the multiple regression analysis,the accuracy of prediction of this method is improved by a big margin with correlation coefficient going up from0.7886to0.9943and relative error lowering to2.19%.