Scientists are changing the way they assess and plan for natural hazards. Many experts are now arguing for an approach that identifies the best practical solutions to cope with hazards, in preference to what they see as a quest for increasingly accurate predictions and an unrealistic emphasis on eradicating uncertainty. The chaotic nature of seismic activity makes it impossible to predict an earthquake with sufficient time for people to leave their homes, for example. Instead, experts are moving toward long-term earthquake forecasts to assess whether buildings should be strengthened. Systems experts look at the likelihood of hazards and the balance the costs and benefits of interventions on our fundamental needs (i.e. water, health, etc.). They share calculations of how society’s finite resources should be spent for use in policy decisions. But the systems approach, which urges action based on risk assessment, is being constrained by demands for better predictions. At the center is a gap between what we worry about and what will actually change outcomes. Bangladesh, for example, is badly affected by changing climate, and some scientists argue that instead of refining a 50-year climate forecast, efforts should focus on calculating when people should abandon their land in floods, to move seeds and tools to higher ground. This session discusses how we decide what to prepare for. Can science and policy be aligned on these issues or are they in perpetual tension?