Our world is currently threatened by digital viruses, such as email viruses and mobile viruses. These viruses are mainly activated by users� operations. Therefore, it�s important for us to understand the pattern of user�s operational behaviors and estimate the effect of such behaviors on virus propagation. This paper first reveals the statistical characteristics of human behaviors, especially the email-checking intervals of the same user based on the Enron email dataset. After that, we analyze the effect of human operational behaviors and network topologies on virus propagation in a human-oriented virus propagation model. The empirical results from real dataset show that the waiting intervals of each user to check mailbox follow a long-tail distribution. Combining this finding, our experiments accurately describe the process of email-virus propagation. The results show that viruses can fast spread in a network if the email-checking intervals follow a long-tail distribution with a higher power-law exponent. Meanwhile, our results find that the infected nodes with the highest-degree may speed up the virus propagation through analyzing the effects of network structure on virus propagation.