BACKGROUND: Tuberculosis (TB) has overtaken HIV as the biggest infectious disease killer, with the majority of deaths occurring in sub-Saharan Africa. However it is unknown how differences in bacterial load alter host immune profiles in the sputum and blood of TB patients. METHODS: 16S ribosomal RNA analysis was used to determine bacterial load in sputum samples obtained from 173 patients with active TB (57 pre-treatment and 116 post-treatment). Host analyte concentrations in sputum and Mycobacterium tuberculosis (Mtb) antigen stimulated whole blood assay supernatants were analysed using multiplex cytokine arrays. RESULTS: Multiple logistic regression adjusting for age, sex and HIV status showed highly significant correlation of bacterial load with IL1β, IL2, IL1RA, IL4, IL6, IL8, IL9, IL15, IL17, EOTAX, FGF, IFN-γ, GCSF, MCP1, M1P1α, M1P1β, PDGF, TNFα, VEGF in sputum. With increasing time on treatment, FGF levels in sputum displayed the most significant inverse correlation with reduction in bacterial load. CONCLUSIONS: We show that differences in bacterial load correlates with changes in several host biomarkers. These findings have implications for development of tests for TB diagnosis and treatment response.