Rationale: A human model to better understand tuberculosis immunopathogenesis and facilitate vaccine development is urgently needed.Objectives: We evaluated the feasibility, safety, and immunogenicity of live bacillus Calmette-Guérin (BCG) in a lung-oriented controlled human infection model.Methods: We recruited 106 healthy South African participants with varying degrees of tuberculosis susceptibility. Live BCG, sterile PPD, and saline were bronchoscopically instilled into separate lung segments (n = 65). A control group (n = 34) underwent a single bronchoscopy without challenge. The primary outcome was safety. Cellular and antibody immune signatures were identified in BAL before and 3 days after challenge using flow cytometry, ELISA, RNA sequencing, and mass spectrometry.Measurements and Main Results: The frequency of adverse events was low (9.4%; n = 10), similar in the challenge versus control groups (P = 0.8), and all adverse events were mild and managed conservatively in an outpatient setting. The optimal PPD and BCG dose was 0.5 TU and 104 cfu, respectively, based on changes in BAL cellular profiles (P = 0.02) and antibody responses (P = 0.01) at incremental doses before versus after challenge. At 104 versus 103 cfu BCG, there was a significant increase in number of differentially expressed genes (367 vs. 3; P < 0.001) and dysregulated proteins (64 vs. 0; P < 0.001). Immune responses were highly setting specific (in vitro vs. in vivo) and compartment specific (BAL vs. blood) and localized to the challenged lung segments.Conclusions: A lung-oriented mycobacterial controlled human infection model using live BCG and PPD is feasible and safe. These data inform the study of tuberculosis immunopathogenesis and strategies for evaluation and development of tuberculosis vaccine candidates.