It has been estimated that every human being carries ~20 rare “natural human gene knockouts”-DNA variants in protein-coding regions of the genome that partially or completely inactivate gene products. When coupled with information about clinical phenotypes, these natural human gene knockouts can illuminate biological function of geness and further our understanding of diseass. In select cases, “protective” gene-inactivating mutations may guide the way to new therapeutic targets or confirm existing targets. With the advent of next generation sequencing facilitating genomics on massive scales, properly powered studies of human knockouts are finally possible. Here, we describe an integrated approach to sequencing-based discovery of naturally occurring human knockouts that spans genetic trait architectures, from small collections of highly related individuals with extreme phenotypic traits, to geographically- and reproductively-isolated “founder” populations harboring frequent instances of highly impactful alleles, to large scale sampling of outbred populations with rich “real-world” phenotypic data captured in electronic medical records. In collaboration with the Geisinger Health System and academic groups worldwide, we report early findings from whole exome sequencing of over 100,000 individuals spanning these population architectures and study designs. We find that the majority of genes encoding drug targets harbor mutations that are predicted to partially or completely inactivate their gene products. We highlight examples of protective and harmful clinical associations with inactivating mutations in these genes that support and invalidate therapeutic targets. We also describe exome-wide scans for gene-inactivating mutations that nominate novel candidate genes and targets in multiple clinical traits and diseases. These early insights suggest that large-scale discovery of human gene knockouts will contribute to the next wave of drug target discovery and validation.