Identification of ballistic missiles contains many aspects, and the identification of launching symptom is an important part of it, which determines the subsequent detection probability of ballistic missile by early warning satellite and the warming time.In this paper, a synthetic method for identification of ballistic missile launching symptom is proposed based on Dynamic Bayesian Network (DBN).Firstly, four kinds of symptom features are selected and the conditional probability of each feature is given.Then the DBN model of symptom identification is established and the identification process is designed.At last, it is verified by simulation that the model is much better and more robust than the method using single feature.The study shows that the method can fuse multiple kinds of features at different time, and give a scientific and reasonable symptom identification result.