A neural network-based residual flight time prediction method is designed for the dynamic target allocation of anti-aircraft missiles.This method not only takes into account the relative missile-to-target distance and the missile motion state，but also considers the influence of the motion state of the enemy target on the prediction results，thereby improving the prediction accuracy.A target allocation model，which comprehensively considers the distance advantage，angle advantage，and residual flight time，is established in combination with the designed method.An auction algorithm is adopted to solve this model，by which an overall optimal target allocation scheme is achieved.The simulation results show that the prediction error of the neural network model is within 0.4 s，which is lower than that of the classical method.The computation time for reallocation with the auction algorithm can meet the real-time requirements of the system.