There is rising evidence of an inverse association be-tween chronic diseases and diets characterized byrich fruit and vegetable consumption. Dietary com-ponents may act directly or indirectly on the humangenome and modulate multiple processes involvedin disease risk and disease progression. However,there is currently no exhaustive resource on thehealth benefits associated to specific dietary inter-ventions, or a resource covering the broad molecu-lar content of food. Here we present the first releaseof NutriChem, available athttp://cbs.dtu.dk/services/NutriChem-1.0, a database generated by text miningof 21 million MEDLINE abstracts for information thatlinks plant-based foods with their small moleculecomponents and human disease phenotypes. Nu-triChem contains text-mined data for 18478 pairs of1772 plant-based foods and 7898 phytochemicals,and 6242 pairs of 1066 plant-based foods and 751diseases. In addition, it includes predicted associa-tions for 548 phytochemicals and 252 diseases. Tothe best of our knowledge this database is the onlyresource linking the chemical space of plant-basedfoods with human disease phenotypes and providesa foundation for understanding mechanistically theconsequences of eating behaviors on health.