As the existing network security situation prediction is restricted to its single information source and poor real-time property,a new prediction method fully considering the variation of network security situation is proposed on the basis of time series analysis.In this method,a series of hidden Markov models are constructed to predict the security situation for hosts according to the front and back dependence,and then to predict the trend of network se- curity at the next moment by fully using multi-source heterogeneous information.Moreover,the network security si- tuation at the next moment is quantitatively calculated from all hosts in the network.Thus,the change law and de- velopment direction of network security situation can be indirectly predicted.Experimental results in real network environments show that the proposed prediction method of network security situation is feasible and effective.