Tuberculosis (TB) remains the deadliest infectious disease1, and the widely used Bacillus Calmette-Guérin (BCG) vaccine fails to curb the epidemic. An improved vaccination strategy could provide a cost-effective intervention to break the transmission cycle and prevent antimicrobial resistance2,3. Limited knowledge of the host responses critically involved in protective immunity hampers the development of improved TB vaccination regimens. Therefore, assessment of new strategies in preclinical models to select the best candidate vaccines before clinical vaccine testing remains indispensable. We have previously established in rhesus macaques (Macaca mulatta) that pulmonary mucosal BCG delivery reduces TB disease where standard intradermal injection fails4,5. Here, we show that pulmonary BCG prevents infection by using a repeated limiting-dose Mycobacterium tuberculosis challenge model and identify polyfunctional T-helper type 17 (TH17) cells, interleukin-10 and immunoglobulin A as correlates of local protective immunity. These findings warrant further research into mucosal immunization strategies and their translation to clinical application to more effectively prevent the spread of TB.