BACKGROUND: BCG is used widely as the sole licensed vaccine against tuberculosis, but it has variable efficacy and the reasons for this are still unclear. No reliable biomarkers to predict future protection against, or acquisition of, TB infection following immunisation have been identified. Lessons from BCG could be valuable in the development of effective tuberculosis vaccines. OBJECTIVES: Within the Entebbe Mother and Baby Study birth cohort in Uganda, infants received BCG at birth. We investigated factors associated with latent tuberculosis infection (LTBI) and with cytokine response to mycobacterial antigen at age five years. We also investigated whether cytokine responses at one year were associated with LTBI at five years of age. METHODS: Blood samples from age one and five years were stimulated using crude culture filtrates of Mycobacterium tuberculosis in a six-day whole blood assay. IFN-γ, IL-5, IL-13 and IL-10 production was measured. LTBI at five years was determined using T-SPOT.TB(®) assay. Associations with LTBI at five years were assessed using multivariable logistic regression. Multiple linear regression with bootstrapping was used to determine factors associated with cytokine responses at age five years. RESULTS: LTBI prevalence was 9% at age five years. Only urban residence and history of TB contact/disease were positively associated with LTBI. BCG vaccine strain, LTBI, HIV infection, asymptomatic malaria, growth z-scores, childhood anthelminthic treatment and maternal BCG scar were associated with cytokine responses at age five. Cytokine responses at one year were not associated with acquisition of LTBI by five years of age. CONCLUSION: Although multiple factors influenced anti-myocbacterial immune responses at age five, factors likely to be associated with exposure to infectious cases (history of household contact, and urban residence) dominated the risk of LTBI.