The lack of knowledge about the relationship between tumor genotypes and therapeutic responses remains one of the most critical gaps in enabling the effective use of cancer therapies. Here we couple a multiplexed and quantitative experimental platform with robust statistical methods to enable pharmacogenomic mapping of lung cancer treatment responses in vivo. The complex map of genotype-specific treatment responses uncovered that over 20% of possible interactions show significant resistance or sensitivity. Known and novel interactions were identified, and one of these interactions, the resistance of KEAP1 mutant lung tumors to platinum therapy, was validated using a large patient response dataset. These results highlight the broad impact of tumor suppressor genotype on treatment responses and define a strategy to identify the determinants of precision therapies.