Bacille Calmette-Guerin (BCG) vaccination for the prevention of tuberculosis has been used in humans since 1921. Furthermore, for >60 years it has been possible to separate BCG strains (defined here as a BCG vaccine maintained in a particular laboratory and used in a particulartrial or set of trials) on the basis of in vitro and in vivo tests. Investigators have concluded that differences in the BCG strains used in efficacy trials on humans may be responsible forthe widerange in levels of protection from tuberculosis reported in those trials. We review the development of the separate strains used in the trials included in a recent meta-analysis and examine data for and against the protective efficacy of different BCG strains. The difficulties incorrelating results of in vitro and in vivo tests with protective efficacy in humans are discussed. Thelimited data available from human studiessuggest that the BCG strain used for vaccinationis not a significant determinant of the overall efficacy in the prevention of tuberculosis.