The outcome of Mycobacterium tuberculosis (Mtb) infection and the immunological response to the Bacille Calmette Guerin (BCG) vaccine are highly variable in humans. Deciphering the relative importance of host genetics, environment, and vaccine preparation on BCG efficacy has proven difficult in natural populations. We developed a model system that captures the breadth of immunological responses observed in outbred individuals, which can be used to understand the contribution of host genetics to vaccine efficacy. This system employs a panel of highly-diverse inbred mouse strains, consisting of the founders and recombinant progeny of the Collaborative Cross. Unlike natural populations, the structure of this panel allows the serial evaluation of genetically-identical individuals and quantification of genotype-specific effects of interventions such as vaccination. When analyzed in the aggregate, our panel resembled natural populations in several important respects; the animals displayed a broad range of Mtb susceptibility, varied in their immunological response to infection, and were not durably protected by BCG vaccination. However, when analyzed at the genotype level, we found that these phenotypic differences were heritable. Mtb susceptibility varied between lines, from extreme sensitivity to progressive Mtb clearance. Similarly, only a minority of the genotypes was protected by vaccination. BCG efficacy was genetically separable from susceptibility, and the lack of efficacy in the aggregate analysis was driven by nonresponsive lines that mounted a qualitatively distinct response to infection. These observations support an important role for host genetic diversity in determining BCG efficacy, and provide a new resource to rationally develop more broadly efficacious vaccines.