The paper takes stock of the debate about the so-called warning-response-gap regarding armed conflict within states. It argues that while the existing literature has focused strongly on "better prediction," it has neglected the analysis of the conditions under which warnings are being noticed, accepted, prioritized and responded to by policy-makers. This has led to a simplistic understanding of how communicative, cognitive and political processes involving a range of actors can influence both the perception as well as the response to warnings. The paper also criticizes that many normative judgments about the desirability of preventive action are suffering from hindsight bias and insufficient attention to balancing problems related to risk substitution, opportunity costs and moral hazard. In response to these deficits, the paper puts forward a modified model of warning as a persuasive process. It can help us to ascertain under what circumstances warning succeed in overcoming cognitive and political barriers to preventive action and to help establishing benchmarks for assessing success and failure from a normative perspective.