Food composition databases are used to quantify dietary intake across a range of applications. In Australia, the 5,740-item AUSNUT 2011-13 database (1) is commonly used in human research. Nutrient information in this comprehensive database is derived from chemical analysis, nutrient imputation and borrowing of nutrient information from similar foods or other databases. Chemical analysis of foods is the most accurate, however this process is labour intensive and expensive, which is a factor limiting expansion of the Australian foods database (2). Other databases such as United States Department of Agriculture (USDA) SR Legacy Foods (3) include more comprehensive data for some nutrients such as all amino acids (AA), while the AUSNUT database only has complete data for tryptophan. The aim of this work was to assess the validity of imputing nutrient values from the USDA database to reduce missing data in AUSNUT 2011-13, with a focus on AA data. Whole foods in AUSNUT were matched to USDA Legacy equivalents by name, prioritising high-protein foods. The nutrient composition of matched foods were compared using food vector analysis including total protein and tryptophan content. Where available, AA content from the USDA database was compared to analysed foods in the Australian Food Composition Database (AFCD) (4) to assess the similarity of content and suitability of using US values in an Australian context. Of 2,090 whole foods (without a recipe file) in AUSNUT, 18% (n = 385) were directly matched to an USDA equivalent based on matched name/food to test the feasibility of methods. Matched whole foods included a range of meats, dairy products, grains, fruits and vegetables. Food vector ratio analysis revealed that 45% (n = 172) of foods were within 10% of USDA values for protein and tryptophan content. Eighty-three matched foods were also successfully paired with their equivalent in AFCD database, allowing the comparison of 16 AAs between US and Australia. Arginine values were most similar between databases, with 72% of foods within ± 30% similarity. A range of fish, eggs and vegetables were most similar in AA content with an average of 64% of foods within ± 30% similarity across all available AAs. This analysis indicates the feasibility of AA data imputation. Next, machine learning will be employed to further populate the Australian food composition database with AAs where chemical testing is not possible to objectively determine AA contents of foods. This work demonstrates that AA contents of foods from an international database have adequate similarity to be used for imputation of nutrient values to reduce missing data within the Australian database. This methodological approach used for borrowing nutrient values could be implemented across various nutrients and contribute to the development of more comprehensive databases for use in human research.