ABSTRACT Millions of Americans have or will develop dementia, which has potentially dire implications for them, their families, and the Medicare program. Beneficiaries with dementia already account for an estimated one in five dollars of Medicare spending, yet the Medicare program has struggled to meet their complex medical and supportive needs. Researchers' efforts to describe and improve the care Medicare beneficiaries with dementia receive have been hampered by the difficulty of identifying those with dementia or cognitive impairments in claims data, yet such data are currently the only data on Medicare beneficiaries that are available on a large scale. Prior studies suggest that many beneficiaries have undetected or undiagnosed disease, while others with intact cognition carry a false claims-based diagnosis. Accordingly, our first aim is to leverage linked survey and claims data to predict the probability of dementia and its severity, according to a validated, survey-based reference standard. Better identification of those with dementia is a prerequisite to obtain a more accurate national picture of the care for this important population. The two data sources are also complementary; survey data help assess under-diagnosis and the probability that a claim-based diagnosis truly represents dementia, and claims data supports the examination of Medicare programs at scale. We will then describe the sorting of Medicare beneficiaries across available plan options: Traditional Medicare (TM), Medicare Advantage (MA), and within MA, Special Needs Plans (SNPs). This work builds on our existing P01 research on choice, which demonstrated that beneficiaries with less cognitive ability were more likely to choose a dominated plan. In this project, we will examine sorting with respect to options potentially advantageous for those with dementia, e.g., SNPs. Finally, we will examine dementia care and outcomes in the changing Medicare program as both payments for dementia beneficiaries and benefit flexibility increase. For example, MA plans will receive substantially larger payments for enrolled beneficiaries with dementia starting in 2020 through risk adjustment formula changes, which coincides with new flexibility in offering LTSS because of the CHRONIC Care Act. Additionally, reimbursement of skilled nursing facilities will change in late 2019 to weigh cognitive deficits more and therapy minutes less. Our study will provide critically needed evidence about the role of incentives and program flexibility for beneficiaries with dementia receiving care within Medicare in an era of change.