Background: Metabolomics study provides an opportunity to identify novel molecular determinants of altered cognitive function. Methods: During 2013 to 2016 Bogalusa Heart Study (BHS) visit, 1,177 participants underwent untargeted, ultrahigh performance liquid chromatography-tandem mass spectroscopy metabolomics profiling. Global cognition and five cognition domains were also assessed. The cross-sectional associations of single metabolites with cognition were tested using multiple linear regression models. Weighted correlation network analysis was used to examine the covariable-adjusted correlations of modules of co-abundant metabolites with cognition. Analyses were conducted in the overall sample and according to both ethnicity and sex. Results: Five known metabolites and two metabolite modules robustly associated with cognition across overall and stratified analyses. Two metabolites were from lipid sub-pathways including fatty acid metabolism [9-hydroxystearate; minimum P-value (min-P)=1.11x10(-5)], and primary bile acid metabolism (glyco-alpha-muricholate; min-P=4.10x10(-5)). One metabolite from the glycogen metabolism sub-pathway (maltose; min-P=9.77x10(-6)), one from the polyamine metabolism sub-pathway (N-acetyl-isoputreanine; min-P=1.03x10(-5)), and one from the purine metabolism sub-pathway (7-methylguanine; min-P=1.19x10(-5)) were also identified. Two metabolite modules reflecting bile acid metabolism and androgenic steroids correlated with cognition (min-P=5.00x10(-4) and 3.00x10(-3), respectively). Conclusion: The novel associations of 5 known metabolites and 2 metabolite modules with cognition provide insights into the physiological mechanisms regulating cognitive function.