Background: Increasing adropin levels in old mice improves cognition. Adropin expression in human brain samples correlates with pathways affecting cognition. Whether circulating adropin levels correlate with cognition in advanced age is not known. Methods: Plasma adropin concentrations were measured in 452 participants of the Multidomain Alzheimer Preventive Trial (MAPT; ClinicalTrials.gov Identifier,NCT00672685; mean age 75.8y, SD=4.5y, 60.2% female). Cognitive ability was assessed using a composite cognitive score (CCS). Relationships between plasma adropin concentrations and changes in CCS (ΔCCS) were examined using regression modelling or grouping into tertiles ranked by adropin values. Results: Plasma adropin levels correlated inversely with ΔCCS. Between adropin tertiles, ΔCCS adjusted for age, time between baseline and final visit, and baseline CCS was significantly different (estimated marginal mean ± SE of ΔCCS for the 1st, 2nd, and 3rd adropin tertiles, -0.299 ± 0.057; -0.252 ± 0.057; -0.107 ± 0.058; n=151,152, and 149, respectively; P<0.05 between 1st and 3rd adropin tertile). Controlling for years of education and medications did not affect the result, while protection was also observed in carriers of the APOE ϵ4 allele. Plasma Aβ42/40 ratio also differed significantly between tertiles (P=0.012), with a higher ratio in the 3rd adropin tertile. Interpretation: This study found an inverse relationship between plasma adropin concentrations and cognitive decline in community-dwelling older adults. Further studies are needed to determine the underlying causes of the relationship, and whether raising plasma adropin concentrations delays or prevents cognitive decline.