Food composition data have a fundamental function in studies on nutrition, health, and agriculture, among others. Many factors affect the nutrient content of food, and for this reason, it is essential to have updated and reliable data on the composition of the main foods consumed. The objective of this work was to develop a food composition database (FCDB) that compiles the composition of native foods of Latin America, mainly grains/seeds, tubers, and derivatives. An interdisciplinary work group of compilers was formed. A search of various sources was carried out (scientific publications, laboratory and technical reports, and theses), and a total of 78 publications were collected. For compilation, a form composed of eight worksheets was prepared. The initial sheet contains general data and food identification; the remaining ones contain information on the proximal composition, amino acids, fatty acids, vitamins, and minerals. Each section has an evaluation of data quality, which determines whether it will be included in the FCDB or not. After an exhaustive analysis based on compliance with the minimum requirements previously established, 58 publications and laboratory reports were selected. The main reason for rejection was the lack of moisture information (60%), followed by low data quality (30%). Information is available on the composition of at least 26 grains and derived products (i.e., quinoa, amaranth, and kañiwa) and five tubers and roots (Andean potatoes and ocas), which are currently being uploaded to the website (http://insibio.org.ar, accessed on 16 December 2021) for user availability. This database will provide information on the composition of regional foods generated and compiled using international standards.