Risk models are built on the best available data—but the best data are often less than ideal. It is only when a disaster occurs that we can retrospectively assess how accurately a risk model predicted the extent and magnitude of disaster impacts. Models sometimes surprise us with their accuracy, but more often, they over- or underestimate the scale of the disaster. Postdisaster forensics offers an opportunity for determining why a risk model has failed, but in our experience this information is not being effectively utilized to improve risk models.