A typical U.S. homebuyer's understanding of whether a property faces flood risk is based on whether the property is located inside the National Flood Insurance Program's (NFIP) Special Flood Hazard Area (SFHA). The SFHA boundary, however, may bias homebuyers’ perceptions of flood risk relative to unobserved true risk because the SFHA is an incomplete, and sometimes inaccurate, representation of flood hazards. Using a national dataset of property transactions, flood hazard data from the First Street Foundation, state-level measures of flood disclosure laws, and a spatially restricted triple-difference design, we distinguish capitalization effects of policy-driven (SFHA) versus quasi-objective (First Street) indicators of flood hazard. We identify these effects by assessing thousands of local spatial interactions between property SFHA designations, measures of quasi-objective flood hazard, and being in a state that mandates flood history disclosures. Being inside the SFHA generates a risk discount, but the signal is muted relative to underlying hazard exposure. Further, the SFHA signal can result in inefficient discounts for properties erroneously mapped in the SFHA. Disclosure requirements about flood history accentuate price effects for hazardous properties and result in more adequate risk internalization. In the absence of disclosures, however, we find either no or a weak risk signal for houses outside the SFHA that face flood hazard. Our results highlight potential benefits of updating SFHA boundaries to include all houses that may experience flooding, and argue in favor of requiring flood-related disclosures from sellers to improve the market's ability to internalize flood risk.