Lipid nanoparticles (LNPs) are formulated using unmodified cholesterol. However, cholesterol is naturally esterified and oxidized in vivo, and these cholesterol variants are differentially trafficked in vivo via lipoproteins including LDL and VLDL. We hypothesized that incorporating the same cholesterol variants into LNPs - which can be structurally similar to LDL and VLDL – would alter nanoparticle targeting in vivo. To test this hypothesis, we quantified how >100 LNPs made with 6 cholesterol variants delivered DNA barcodes to 18 cell types in wildtype, LDL R−/−, and VLDLR−/− mice that were both age-matched and female. By analyzing ~2,000 in vivo drug delivery data points, we found that LNPs formulated with esterified cholesterol delivered nucleic acids more efficiently than LNPs formulated with regular or oxidized cholesterol when compared across all tested cell types in the mouse. We also identified an LNP containing cholesteryl oleate that efficiently delivered siRNA and sgRNA to liver endothelial cells in vivo. Delivery was as - or more - efficient than the same LNP made with unmodified cholesterol. Moreover, delivery to liver endothelial cells was 3X more efficient than delivery to hepatocytes, distinguishing this oleate LNP from hepatocyte-targeting LNPs. RNA delivery can be improved by rationally selecting cholesterol variants, allowing optimization of nanoparticle targeting.