Here's a paper dealing with a frequently worrisome issue:

Shrestha, P. M., Nevin, K. P., Shrestha, M., & Lovley, D. R. (2013). When Is a Microbial Culture “Pure”? Persistent Cryptic Contaminant Escapes Detection Even with Deep Genome Sequencing. mBio, 4(2). doi:10.1128/mBio.00591-12
Microbiologists, like most biologists, like to have neatly organized categories and taxonomies available at all times. We work with individual species, specific strains of those species, and often mutants of those strains. If we grow our chosen organisms and they don't behave as expected, we generally assume that contamination occurred somewhere along the line or that a new mutation transpired. The usual preventative measure is simply to ensure that cultures start from single cells of the desired strain.

Shrestha et al. seem to have had some issues with culture contamination. This group works with Geobacter sulfurreducens strain DL1 in studies of microbial fuel cells. As they were attempting to isolate G. sulfurreducens strains with mutations enabling them to be more electrically conductive, they managed to isolate a strain later named KN400. This strain was great at conducting current but clearly wasn't just a mutant of DL1: sequencing showed that the new strain likely differed by more than 27,000 SNP's, or far more mutations than any lab-based methods of selection were likely to create.

The authors sought to clarify the origins of G. sulfurreducens strain KN400. Next-gen sequencing showed that a sequence specific to KN400 could be found in the original DL1 cultures, though out of more than 107 sequence copies, fewer than 300 were the KN400 sequence. The new strain wasn't a new mutant at all. Subsequent growth studies found that even repeated rounds of restreaking cells on solid medium still yielded colonies with minute but detectable levels of the cryptic KN400 strain. Shrestha et al. suggest that their intense electrical conductivity-based selective pressures may have been the only reason they were able to isolate KN400 at all.

It's still not clear how G. sulfurreducens KN400 lives alongside the DL1 strain so well. My guess is that KN400 is actually an instance of a separate but phylogenetically related chromosome copy acting as a mobile genetic element. Cultures of the DL1 strain consistently maintain KN400 cells at low levels, so some variety of fitness benefit must be conferred to the entire culture or we'd expect the minority cells to just get selected against. There's clearly something strange going on with this bacterial species but it may be an ongoing process.