As Hurricane Katrina and other recent major floods have illustrated, flooding causes multiple deaths and billions of dollars of property damage in the United States each year. Over the long term, flood damages are mitigated through flood management, which relies on estimates of the climatological risk of extreme precipitation and runoff. Estimates of flood risk are highly uncertain for many reasons, including societal and land use change, climate variability and change, and fundamental difficulties in estimating the climatological risk of extreme weather events. This uncertainty has substantial impacts on flood management decisions, public safety, and costs. By analyzing two cases in which policy makers were presented with conflicting scientific estimates of flood risk, we examine the challenges in estimating the climatological risk of extreme precipitation and the interplay among scientific uncertainty, flood management decision-making, and societal outcomes. The analysis considers the differing perspectives of various participants in the flood management process, illustrating how uncertainties attributable to scientific issues interact with values issues. The results indicate that, to achieve desired safety levels while negotiating difficult trade-offs between diverse interests, decision makers need to know how reliable their scientific information is, and scientists must acknowledge the interplay between scientific uncertainty and values. The lessons from this study of flood management may also apply to other areas, such as climate change, in which decisions involve scientific estimates at the interface between climate and extreme weather.