The U.S. Geological Survey is developing national water-use estimates as part of a project to support water resources management in the United States. Models support a nationally consistent approach to estimating water-use, are repeatable and reusable for their stated purpose which supports efficient updates to the results and have capabilities that allow users to generate forecasts of future water-use based on dynamic conditions. In this data release water-use refers to water extracted by domestic self-supplied consumers. This data release contains data used in a machine learning model that estimates the counts and distribution of households with a self-supplied water source (self-supplied households) within the conterminous United States for decadal year 2020. This data release also contains the associated scripts used to constrain, train, and validate the model and the scripts used to make estimations of self-supplied household counts, distribution, and water use from model results. Results are stored in raster format as monthly indoor and outdoor self-supplied water use and annual self-supplied water use.