Abstract: In view of fixed learning rate, slow convergence and easy to fall into local minimum problems in the traditional error back propagation (BP) neural network algorithm, the BP neural network method based on credit is proposed. During the layers of weights and threshold adjustment, this method makes the learning rate change continually, jump out of local minima area and eliminate oscillation by introducing contributions of each weights to errors, combining with the momentum coefficient method, which speeds up the convergence rate and improve the efficiency of the network learning. Simulation results show that the improved BP neural network improves network learing speed, in the same time, convergence is better than traditional BP neural network.