AbstractBackgroundStudies on the health effect of Supplemental Nutrition Assistance Program (SNAP) on the cognitive health of older adults are scarce. In this study, we examined the associations of SNAP use and memory function trajectories among US older adults.MethodOur sample included 3,555 SNAPâeligible participants (aged 50+) from the 1996 Health and Retirement Study survey, among whom 559 were SNAP users and 2,996 were nonâusers. SNAP eligibility was constructed based on household income and assets compared to the federal criterion. Memory function was assessed biennially from 1996 through 2016. To account for preâexisting differences in characteristics and variations in survey attrition between SNAP users and nonâusers, we modeled the probability of SNAP use and of attrition using preâSNAP exposure (i.e., 1994) demographic and health covariates. We then modeled trajectories of memory function across SNAP use using linear mixed effect models weighted by the inverse probability of treatment (i.e. SNAP use) and of attrition. We also crossâexamined the results in a propensity score (PS) matched sample (N = 1,050).ResultSNAP users in 1996 had lower socioeconomic status and worse health at baseline than SNAP nonâusers, and were more likely to be lost to followâup. Results using IPâweights suggested SNAP users had worse memory scores at baseline but slower memory decline compared with nonâusers (annual decline rate is â0.038 standard units [95%CI = â0.044, â0.033] for nonâusers and â0.043 [95%CI = â0.046, â0.040] for users). Results from the PSâmatched sample also showed SNAP users had slower memory declines compared with nonâusers (annual decline rate was â0.052 units [95%CI = â0.055, â0.050] for nonâusers and â0.048 units [95%CI = â0.051, â0.046] for users).ConclusionAfter accounting for preâexisting differences between eligible SNAP users and nonâusers as well as differential loss to followâup, our findings suggested that SNAP participation is associated with slower memory decline among eligible older adults.