RNA- and DNA-based drugs can specifically turn any gene on or off. However, whether the drug is made out of siRNA, mRNA, Zinc Finger Nucleases, CRISPR, or others types of DNA / RNA, all genetic drugs are limited by one universal problem: drug delivery. When injected, DNA and RNA drugs tend to accumulate in the liver. Non-liver delivery remains a significant problem. Engineers and chemists have designed thousands of chem. distinct nanoparticles to deliver these drugs to target tissues. After synthesizing these chem. diverse nanoparticle 'libraries', the nanoparticles are typically screened in vitro. However, in vitro conditions typically lack an immune system, kidney, spleen, pulsatile blood flow, and other factors that affect the nanoparticle in vivo. For example, we recently compared how 400 different nanoparticles delivered genetic drugs in vitro and in vivo; we found no correlation. We therefore reasoned that a method to rapidly screen thousands of nanoparticles directly in vivo would allow us to rapidly and efficiently discover nanoparticles with novel tropisms. To this end, we developed a series of increasingly advanced DNA barcoded nanoparticle systems. Here we report three such systems that can (1) measure hundreds in vivo nanoparticle biodistribution with 100,000,000x more sensitivity than fluorescence, (2) quantify how hundreds of nanoparticles functionally deliver mRNA to dozens of cell types in a single mouse, or (3) quantify how hundreds of nanoparticles functionally deliver siRNA to dozens of cell types in a single mouse. Finally, we describe a bioinformatics pipeline to iteratively use these large datasets in order to 'evolve' LNPs with new tropisms. Using these new assays and new anal. pipeline, we have identified LNPs that deliver different many types of therapeutic RNAs to new cell types in vivo.