Identification of metabolites in humans remains challenging. Here, we present synthesis-based reverse metabolomics as a strategy for structure elucidation that aims to also find phenotypic associations. In this approach, MS/MS spectra are acquired from newly synthesized compounds, and searched against public metabolomics data to uncover their phenotypic associations. To demonstrate the concept, we used combinatorial amide coupling reactions to synthesize an array of amino acid conjugated bile acids. A total of 16,587 spectral matches were found for 145 amidates synthesized, representing the single largest expansion of structures in the 170+ year history of bile acids. Furthermore, some new bile acids were associated with health-related phenotypes such as inflammatory bowel diseases and obesity. Using independent human cohorts for validation revealed that some cholic and chenodeoxycholic acid conjugates were elevated in Crohn’s disease, of which some were also potent PXR agonists. Bacteria belonging to bifidobacterium, clostridium, and enterococci genera were the main producers of these new conjugated bile acids. Because searching repositories with MS/MS spectra has only recently become possible, this synthesis-based reverse metabolomics approach can now be employed as a general strategy to elucidate structures and discover other new molecules from human and animal ecosystems.