BACKGROUND: Globally, BCG vaccination varies in efficacy and has some non-specific protective effects. Previous studies comparing BCG strains have been small-scale, with few or no immunological outcomes and have compared TB-specific responses only. We aimed to evaluate both specific and non-specific immune responses to different strains of BCG within a large infant cohort and to evaluate further the relationship between BCG strain, scarring and cytokine responses. METHODS: Infants from the Entebbe Mother and Baby Study (ISRCTN32849447) who received BCG-Russia, BCG-Bulgaria or BCG-Denmark at birth, were analysed by BCG strain group. At one year, interferon-gamma (IFN-γ), interleukin (IL)-5, IL-13 and IL-10 responses to mycobacteria-specific antigens (crude culture filtrate proteins and antigen 85) and non-mycobacterial stimuli (tetanus toxoid and phytohaemagglutinin) were measured using ELISA. Cytokine responses, scar frequency, BCG associated adverse event frequency and mortality rates were compared across groups, with adjustments for potential confounders. RESULTS: Both specific and non-specific IFN-γ, IL-13 and IL-10 responses in 1341 infants differed between BCG strain groups including in response to stimulation with tetanus toxoid. BCG-Denmark immunised infants showed the highest cytokine responses. The proportion of infants who scarred differed significantly, with BCG scars occurring in 52.2%, 64.1% and 92.6% of infants immunised with BCG Russia, BCG-Bulgaria and BCG-Denmark, respectively (p<0.001). Scarred infants had higher IFN-γ and IL-13 responses to mycobacterial antigens only than infants without a scar. The BCG-Denmark group had the highest frequency of adverse events (p=0.025). Mortality differences were not significant. CONCLUSIONS: Both specific and non-specific immune responses to the BCG vaccine differ by strain. Scarring after BCG vaccination is also strain-dependent and is associated with higher IFN-γ and IL-13 responses to mycobacterial antigens. The choice of BCG strain may be an important factor and should be evaluated when testing novel vaccine strategies that employ BCG in prime-boost sequences, or as a vector for other vaccine antigens.