Abstract Post-stroke dementia is an important and understudied component of the vascular contributions to cognitive impairment and dementia. Having a stroke approximately doubles the risk of incident dementia for at least a decade afterwards, even after accounting for other vascular risk factors of dementia and the initial effects of the stroke lesion on cognition. Also, silent strokes occur in nearly half of all aging individuals and are associated with dementia. We established in wildtype mice that stroke triggers chronic neuroinflammation in the stroke scar and connected brain regions, and that this causes delayed-onset cognitive decline. In humans, there is neuroinflammation in the stroke scar in about half of all chronic stroke survivors on autopsy, even decades after stroke, suggesting it may play a role in people as well. However, there are no biomarkers that can currently be used in living humans to detect who is at risk of cognitive decline and dementia after stroke. Here we propose to test the hypothesis that inflammation-induced angiogenesis in the stroke and connected regions results in immature leaky vessels that cause blood-brain barrier leakage even very late after stroke. We will recruit 200 participants with chronic stroke and 50 controls at 3 sites (Stanford School of Medicine, Columbia University, and the University of Manchester). We will test an MRI-based imaging biomarker in Aim 1 and ask whether blood-brain barrier permeability is compromised for years after stroke. In Aim 2 we will ask whether a blood biomarker of imbalanced angiogenesis is dysregulated in chronic stroke. For both, we will also look at risk factors for their development and how they relate to stroke size, location, sex, age, and NIHSS. Finally, in Aim 3 we will use both traditional multivariable and machine learning models to ask whether each biomarker separately or together predicts cognitive decline after stroke, and to identify other MRI, blood, and clinical characteristics that are associated. If we are successful, we will establish that there is chronic blood-brain barrier dysfunction after stroke and link it to dysregulated angiogenesis as a potential mechanism. This would be a fundamental change in how post-stroke dementia is conceptualized and would open avenues for novel therapy development. Our predictive models will also be useful to identify stroke survivors at high risk of cognitive decline and/or to select patients for future clinical trials. This will thus help us better understand vascular contributions to cognitive impairment and dementia.