Genetic mutations have been progressively introduced to BCG by repeated serial passage over many decades of its culture and global dissemination. Thus, marked differences exist in the phenotype, antigenicity, reactogenicity, and clinical characteristics of the numerous substrains of BCG currently in use for bladder cancer immunotherapy. These differences influence proposed mycobacterial antitumour mechanisms and toxicity, potentially resulting in variations in clinical efficacy and adverse effects. However, although there is evidence of substrain-related differences in the clinical efficacy of BCG as a tuberculosis vaccine, evidence of an effect on bladder cancer immunotherapy remains elusive, owing to the lack of appropriately powered head-to-head comparative clinical trials, the nonstandardization of BCG manufacture, and variation in treatment protocols--possibly itself a response to underlying substrain differences. Advances in our understanding of mycobacterial genetics, structure and function, and host-pathogen interactions might explain differences in clinical practice and outcomes. These advances are guiding the identification of biomarkers for reactogenicity and efficacy, and the rational design of immunotherapeutic strategies to eliminate the use of live bacilli for bladder cancer therapy.