The laboratory strain H37Rv represents one of the most commonly used strains in the study of Mycobacterium tuberculosis. Despite the apparent stability of the strain, the absence of a selective pressure for virulence factors could lead to the in vitro accumulation of attenuated mutants. To assess this hypothesis, we performed a systematic analysis of individual clones isolated from subcultured M. tuberculosis H37Rv and from a non-subcultured frozen stock. First, we studied two virulence indicators: neutral red staining and content of phthiocerol dimycocerosates (PDIMs). We found that H37Rv formed a mixed population containing wild-type cells, as well as neutral red and PDIM mutants. Then, we compared the global gene expression of 3 isolated clones (which displayed various phenotypes) and the non-subcultured stock, by microarray analysis. This transcriptional profiling confirmed that a significant heterogeneity existed despite, and in addition to, the neutral red and PDIM phenotypes. These results strongly suggest that great caution must be taken in extrapolating data obtained with M. tuberculosis H37Rv grown in vitro, and it would be prudent to study several independent clones to obtain valid conclusions. For this purpose, the neutral red and PDIM phenotypes might be useful indicators of undesired heterogeneity.