For more than two decades, the U.S. National Oceanic and Atmospheric Administration (NOAA) has published a count of weather-related disasters in the United States that it estimates have exceeded one billion dollars (inflation adjusted) in each calendar year starting in 1980. The dataset is widely cited and applied in research, assessment and invoked to justify policy in federal agencies, Congress and by the U.S. President. This paper performs an evaluation of the dataset under criteria of procedure and substance defined under NOAA’s Information Quality and Scientific Integrity policies. The evaluation finds that the “billion dollar disaster” dataset falls short of meeting these criteria. Thus, public claims promoted by NOAA associated with the dataset and its significance are flawed and at times misleading. Specifically, NOAA incorrectly claims that for some types of extreme weather, the dataset demonstrates detection and attribution of changes on climate timescales. Similarly flawed are NOAA’s claims that increasing annual counts of billion dollar disasters are in part a consequence of human caused climate change. NOAA’s claims to have achieved detection and attribution are not supported by any scientific analysis that it has performed. Given the importance and influence of the dataset in science and policy, NOAA should act quickly to address this scientific integrity shortfall.