Abstract: On the basis of analyzing the current network information filtering technologies, a scalable multi-level and multi-policy model of a distributed network information filtering system was described, which combines the functions of the packet analyzing, filtering and forwarding. A fine-granularity information filtering policy was provided, which effectively enhances the information security, increases the speed of data transfer, and decreases the latency. The filtering plug-ins can be conveniently configured and loaded. The model is scalable and open in architecture that can be easily configured and organized in order to meet a variety of requirements and run on the distributed system for gaining better performances. The information processing is transparent to the client hosts and applications so that the side effects to the users are minimal.