Abstract: Runway occupancy time（ROT）is an important factor to measure runway throughput. A method based on BP neural network and runway radar monitoring data is proposed to predict the runway occupancy time of airport arrival flights. Based on BP neural network algorithm to realize prediction. The algorithm uses the airport runway monitoring data in three airports and the quick access recorder（QAR）data such as the speed and acceleration parameters of landing aircraft. These data cover the landing data and taxiing data of more than ten types of aircraft including heavy, medium and light aircrafts. Finally, the weighted average R square value of the regression analysis between the predicted value and the actual observation value of the predicted runway occupation time based on the neural network algorithm is 0.91. The root mean square value of the predicted runway exit distance is 0.95. It shows that the model has good accuracy in predicting the runway occupation time and runway exit distance.