U.S. Latinos are highly diverse, with individuals originating from central and south America, as well as the Caribbean. Latinos are socioeconomically disadvantaged compared to U.S. Whites, and suffer from a higher prevalence of Alzheimer’s disease and related dementia. Genetic determinants of Alzheimer’s disease may be somewhat different in Latinos compared to Whites: in preliminary results, we found that genetic background, in the form of proportion of genome inherited from Amerindian (Native American) ancestors mitigates the detrimental effect of the APOE-𝜖𝜖4 variants on cognitive decline in individuals from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). This suggests heterogeneity in genetically-based mechanisms underlying neurocognitive decline in Latinos. Untargeted metabolomics is a promising approach to develop potential biomarkers for human disease. Previous metabolomics studies identified potential biomarkers for Alzheimer’s disease and for cognitive traits in a cross- sectional manner, however, very little work was done specifically in Latinos, and the genetics and metabolomics have not been integrated for risk prediction, apart from using APOE-𝜖𝜖4. Here, we propose to utilize a dataset including ~1,500 older Latino adults from the HCHS/SOL with untargeted metabolomics measured at a baseline visit, and neurocognitive function estimated in both the baseline and a follow-up visit. Our goals are: (1) develop preliminary biomarkers for aging-related cognitive decline phenotypes (e.g. change in global cognitive function, mild cognitive impairment); (2) study the genetic basis of the association of metabolites with cognitive decline, and study potential heterogeneity by genetic background; and (3) form recommendations and a grant application to support a larger research program utilizing genetics and metabolomics, and potentially other ‘omics, to develop biomarkers of Alzheimer’s disease in diverse populations, and study underlying mechanisms.