Aiming at the fact that the impact of quarantine on network virus propagation has not been considered in existing dynamical models, a network virus paopagation model with dynamical quarantine was proposed on the basis of the simple epidemic SI model. The basic idea of the model is that the suspicious node is quaranteened immediately as soon as the network intrusion detection system finds the suspicious behavior of node. Then, the isolated node is released after a short of period. Furthermore, by using stochastic probability theory, under the assumption that the activity of network nodes follows a Poisson distribution with different degrees of activity, a mathematical expression is obtained for the number of network nodes that may be isolated at any one time. Finally, numerical experiments simulate the change trend of the speed and scale of network virus spread over time with different values of isolation parameters. Simulation results show that quarantine measures can effectively delay the spread of network viruses, and quarantine time has a significant impact on the speed and scale of network viruses. When quarantine time exceeds 0.5 minutes, the scale of network virus spread will be reduced by about 72%.