Systems and methods are described that enable the mitigation of network attacks directed to specific sets of content on a content delivery system. A set of content targeted in the attack may be identified based at least in part on a combination of network addresses to which attacked-related packets are transmitted. Thereafter, the content delivery system may mitigate the attack based on the identified target. For example, where both targeted and non-targeted sets of content are associated with the attacked network addresses, traffic directed to these sets of content may be separated, e.g., in order to reduce the impact of the attack on the non-targeted sets of content or increase the computing resources available to the targeted content. Redirection of traffic may occur using either or both of resolution-based redirection or routing-based redirection.