Large-scale hydrological and water resource models (LHMs) require water storage and release schemes to represent flow regulation by reservoirs. Owing to a lack of observed reservoir operations, state-of-the-art LHMs deploy a generic reservoir scheme that may fail to represent local operating behaviors. Here we introduce a new dataset of bespoke water storage and release policies for 1,930 reservoirs of conterminous United States. The Inferred Storage Targets and Release Functions (ISTARF-CONUS) dataset relies on a new inventory of observed daily reservoir operations (ResOpsUS) to generate reservoir operating rules for 595 data-rich reservoirs. These functions are developed in a standardized form that allows for extrapolation of operating schemes to 1,335 data-scarce reservoirs—leading to the first inventory of empirically derived reservoir operating policies for all large CONUS reservoirs documented in the Global Reservoir and Dams (GRanD) database. Evaluation of the new scheme in daily simulations forced with observed inflow demonstrates substantial and robust improvement for both release and storage relative to the popular Hanasaki method. Performance of the extrapolation approach for data-scarce reservoirs is evaluated with leave-one-out validation and is shown to also offer modest gains on average over Hanasaki. ISTARF-CONUS may be readily adopted in any LHM featuring large reservoirs of the conterminous United States.