Little is known about incidence of vascular and Alzheimer dementias in American Indians. We conducted a large, heterogeneous, population-based, longitudinal cohort study of brain aging in community-dwelling American Indians aged 64-95 years from 11 tribes across 3 states, with neurologic examinations, 1.5T MRI, and extensive cognitive testing. Visit 1 in 2010-2013 (n = 817) and visit 2 in 2017-2019 (n = 403) included all willing, surviving participants. Standardized cognitive tests at both visits included Modified Mini-Mental Status Examination (MMSE), Wechsler Adult Intelligence Scale digit symbol coding (WAIS), Controlled Oral Word Association (COWA), and California Verbal Learning Test short form (CVLT). Test materials added at follow-up included Wide Range Achievement (reading) Test (WRAT) and National Alzheimer Coordinating Center Uniform Data Set cognitive battery (v3 form C2), including Montreal Cognitive Assessment (MoCA). MRI neuroradiologists coded infarcts, hemorrhages, white matter hyperintensities, sulcal atrophy, and ventricle enlargement. The mean time between examinations was 6.7 years (SD 1.1, range 3.8-9.1 years). Years of formal education had modest correlation with WRAT reading score (r = 0.45). Prevalence and incidence (respectively) of infarcts were 32% and 12.8/1,000 person-years (PYs) hemorrhages 6% and 4.4/1000 PY worsening sulci 74% and 19.0/1000 PY worsening ventricle 79% and 30.1/1000 PY worsening leukoaraiosis 44% and 26.1/1000 PY. Linear losses per year in cognitive scores were 0.6% MMSE, 1.2% WAIS, 0.6% COWA, and 2.2% CVLT. The mean MoCA scores were 18.9 (SD 4.3). These are the first data on longitudinal cognitive and imaging changes in American Indians and first reports of Alzheimer disease-related features. The mean scores in MoCA were similar or lower than standard cutoffs used to diagnose dementia in other racial/ethnic groups, suggesting that standardized cognitive tests may not perform well in this population. Test validation, adaptation, and score adjustment are warranted. Years of education were a poor proxy for premorbid function, suggesting novel methods for cognitive score contextualization is also needed in this population. Evaluation of selective survival suggests attrition from death, and frailty should be accounted for in causal analyses. Overall, these data represent a unique opportunity to examine neurology topics of critical importance to an understudied population.