Tuberculosis (TB) remains a leading global cause of morbidity and mortality and an effective new vaccine is urgently needed. A major barrier to the rational development of novel TB vaccines is the lack of a validated immune correlate or biomarker of protection. Mycobacterial Growth Inhibition Assays (MGIAs) provide an unbiased measure of ability to control mycobacterial growth in vitro, and may represent a functional correlate of protection. However, the biological relevance of any potential correlate can only be assessed by determining the association with in vivo protection from either a controlled mycobacterial infection or natural development of TB disease. Our data demonstrate that the direct MGIA using peripheral blood mononuclear cells (PBMC) is measuring a biologically relevant response that correlates with protection from in vivo human BCG infection across two independent cohorts. This is the first report of an MGIA correlating with in vivo protection in the species-of-interest, humans, and furthermore on a per-individual as well as per-group basis. Control of mycobacterial growth in the MGIA is associated with a range of immune parameters measured post-BCG infection in vivo including the IFN-γ ELISpot response, frequency of PPD-specific IFN-γ or TNF-α producing CD4+ T cells and frequency of specific sub-populations of polyfunctional CD4+ T cells. Distinct transcriptomic profiles are associated with good vs. poor mycobacterial control in the MGIA, with good controllers showing enrichment for gene sets associated with antigen processing/presentation and the IL-23 pathway, and poor controllers showing enrichment for hypoxia-related pathways. This study represents an important step toward biologically validating the direct PBMC MGIA for use in TB vaccine development and furthermore demonstrates the utility of this assay in determining relevant immune mechanisms and pathways of protection.