The transmission of malicious information is often accompanied with normal information flows in complex networks. For the intrusion of unknown malicious information, if there is no effective defensive method to suppress its diffusion, the network may be confronted with incredible catastrophe. In this paper, to enhance the security of information diffusion in complex networks, a node trust-value management mechanism (NTVMM) is proposed. Basically, node trust value and threshold are assigned with node importance at the beginning. According to node benefit and loss in information diffusion, the two algorithms, node trust-value updated algorithm (NTVUA) and node trust-threshold updated algorithm (NTTUA), are devised to dynamically update node trust value and trust threshold, respectively. Compared with the existing node quarantine and edge blockage schemes, NTVMM only relies on the trust relationship of nodes to suppress malicious information diffusion with no need for knowing the global infection information of the network. Besides, to further make a trade-off between information security and information loss, NTVMM is associated with the classical information diffusion model to obtain the most appropriate trust threshold. Finally, we devise the network profit function to evaluate the profits brought by different methods. The simulations are respectively carried out in two synthetic complex networks and a real-world complex network. The results of simulation demonstrate that when malicious information invades the network, NTVMM can reduce the number of infected nodes in the network and enhance the security of information diffusion. Moreover, compared with node quarantine and edge blockage schemes, the proposed scheme can obtain a better network profit, and make a trade-off between information security and information loss.