Belowground urban stormwater network (BUSN) data are usually not available to the public at the regional or national scales, hindering predictive understanding of BUSN’s impacts on urban flooding under extreme climates. We derive a national BUSN dataset over the contiguous United States by leveraging a newly developed algorithm based on graph theory and extensively available information such as street networks, landuse, and topography. Additionally, we employ a network-based minimum slope enforcement methodology to ensure hydraulic feasibility of the generated BUSNs. For the convenience of hydrologic modeling, the generated BUSN national dataset is available in a vector format and organized to the 8-digit Hydrologic Unit (HUC8) watersheds. There are two categories of data included in this dataset: 1) network-level information, such as the spatial typology and connectivity between the stormwater pipes; 2) pipe-level information, such as pipe length and slope. Data quality control is also performed to ensure the completeness of BUSN within each HUC8 watershed. We suggest this dataset can be valuable for understanding and modeling urban flooding processes at regional and larger scales.