Food composition databases are a basic tool to assess dietary status of a population and are therefore critical for nutrition policy development. The United States Department of Agriculture (USDA) Agricultural Research Service develops and disseminates widely used food composition databases. The two largest are the Standard Reference (SR) and Survey databases. These two databases have a similar number of foods (6000}7000) but di!erent structures and uses. This paper describes these two databases from the user's perspective. The SR database includes foods used in the United States for which up-to-date nutrient data are available. Foods are listed unprepared (raw) and by the most common preparation methods. The Survey database, used with national surveys of food intake, describes foods in consumer terms. The SR database lists approximately 77 food components, but may not have values for components in all foods. The Survey database lists fewer components (about 50) but has values for these components for every food on the database. Many food composition databases are derived from the SR and Survey databases. They form the backbone for databases on popular software programs and in printed references and are used around the world in creating databases to characterize local food intake.