The bacillus Calmette-Geurin (BCG) family of vaccines currently implemented to prevent tuberculosis (TB) consist of clonal bacterial strains independently shaped by nearly a half-century of evolution. Derived from virulent Mycobacterium bovis, daughter strains of BCG were additionally passaged under the same laboratory conditions that resulted in its original attenuation. Genomic loss of the RDI region has been demonstrated to coincide with attenuation from virulence, while deletions occurring after the loss of RDI are speculated to be responsible for BCG's over-attenuation. To provide a more complete description of their total genomic variation, the genomic content of BCG strains are investigated by Affymetrix GeneChip(TM). Because clinical isolates of M. tuberculosis have previously been characterized via GeneChip(TM) interrogation, analysis permits the comparison of in vivo versus in vitro evolution of M. tuberculosis complex subspecies. The contrast between the two modes of evolution are discussed in its relevance towards TB pathogenicity. (C) 2003 Elsevier Ltd. All rights reserved.