Polygenic risk scores (PRS), which aggregates genetic risk across tens to millions of associated genetic variants, provide a powerful approach to identify individuals at high-risk for common disease. To optimize the clinical utility of PRS in risk prediction, it is critical that we accurately identify the subset of individuals at greatest risk for disease. A major complication of applying PRS to genetically admixed individuals is that PRS distributions are not independent of ancestry. Differences in allele frequencies between contributing continental ancestries groups combined with differences in proportions of continental ancestry can result in significantly different PRS distributions for genetically admixed individuals. We have recently shown that global ancestry calibration of PRS scores, essentially regressing out global ancestry estimates from the PRS, in genetically admixed African Americans can significantly improve prediction performance compared to uncalibrated PRS for traits such as height and diabetes. We hypothesize that ancestry calibration methods that focus on local ancestry, rather than global ancestry proportions, should further improve PRS prediction in genetically admixed individuals. The rationale is simple, global ancestry calibration assumes a common constant adjustment across all included variants in the score when, in fact, there are many variants where the correction should be negligible and many other variants where the correction should be in the opposite direction. We are building a “bootstrap-based” approach using unrelated reference individuals, selected to be representative of broad global genetic diversity, to construct local-ancestry-matched personalized null PRS distributions. Upon constructing a null PRS distribution, an individual’s observed PRS value is “calibrated” by calculating a percentile reflecting how extreme the observed PRS score is compared to the personalized null distribution. The approach is flexible to allow ancestry- and sex-specific allelic effects. We will compare the local ancestry calibration method to uncalibrated PRS and standard global ancestry calibrated PRS.