Filtering the spoofed packets with a false source addresses is the inherent requirement of the trustworthy and secure Internet. Routing based distributed packet filtering is effective, but its effectiveness has no solid theory analysis. In this paper, based on the inter-domain route distribution and the hierarchy of the Internet topology, the study establishes the route distribution tree model and ideal AS graph model using these two models analyze the effectiveness of maximum filtering and semi-maximum filtering. The analysis results verify the former experimental results and figure out the theoretical explanation. Maximum filtering can filter out most spoofed packets. Though it cannot reach 100%, maximum filtering can limit the number of the successful spoofing AS to the average AS path length of the Internet. On the ideal AS graph, semi-maximum filtering has the same effectiveness as the maximum filtering and its storage and computing overhead is much lower than maximum filtering, which provides the theoretical basis to use it in practice. The model-based analysis points out the inherent features of the inter-domain routing based distributed packet filtering, which conduces to design the subsidiary mechanism and the overall deployment in the whole Internet.