The Bayesian theory of flood warning systems awaits further research in three directions: (1) Development of a detailed model and assessment method that would permit one to estimate the disutility function for a community based on survey data; (2) development of comprehensive case studies that would offer guideposts for real-world implementation in design, operation, performance evaluation, and benefit-cost analyses; (3) extension of the theory to probabilistic river stage forecasts that will be available in the future.