A new network security risk assessment method based on hidden Markov model is proposed in order to accurately assess the network security risk in real time. The method is based on Hidden Markov model to model the target network. The security risk of node is quantified by direct risk and indirect risk caused by correlation of the node. Considering the importance of nodes in the network, the overall security risk of the target network is quantified on the basis of node security risk. The experimental results show that the method can quantify the network security risk caused by the correlation and importance of the node, which makes the network security risk assessment results more accurate and credible. Compared with traditional network security risk assessment methods, this method can detect abnormal risk changes in the work more timely ,which can provide the basis for the timely adjustment of the network security defense strategy.