Narrative approaches to analyzing risks seek to identify the variables critical to creating and controlling a risk, then to instantiate them in terms of coherent themes (e.g., organizational failure, strategic surprise). Computational approaches to analyzing risks seek to identify the same critical variables, then to instantiate them in terms of their probability. Disaster risk analysis faces complex, novel processes that strain the capabilities of both approaches. We propose an approach that integrates elements of each, relying on what we call structured scenarios and computable models. It is illustrated by framing the analysis of plans for a possible avian flu pandemic. Copyright Springer Science + Business Media, LLC 2006Scenarios, Uncertainty, Risk analysis, Judgment, Avian flu,