We propose a prospective observational cohort study of 660 hospitalised patients diagnosed with HIV-associated TB with a CD4 count < 350/microlitre and not currently on ART. Participants will be managed according to local guidelines and followed for 12 weeks. We will determine: 1) Cumulative mortality at 12 weeks. 2) Whether quantitative markers of Mycobacterium tuberculosis load correlate with clinical and laboratory markers of sepsis syndrome and whether these are associated with mortality. 3) If severe HIV-TB is associated with high levels of intestinal translocation of bacterial products and viable bacteria and the association with mortality. 4) The proportion of participants with cytomegalovirus viraemia and its association with mortality. 5) Whether immune activation-induced apoptosis and anti-inflammatory signalling (PD-1 pathway and anti-inflammatory cytokines) are higher in participants with fatal outcome. 6) The contribution of paradoxical TB-IRIS to mortality. 7) Wh at proportion of hospitalized HIV-TB patients have subtherapeutic concentrations of antitubercular drugs (INH, rifampicin, PZA) during the critical early period of TB treatment (pharmacokinetic study at 3 days on treatment). Significant findings from this study will allow us to propose novel interventions aimed at reducing mortality in severe HIV-TB and seek funding for RCTs.