Numerous industries and facilities have been using chlorinated ethenes (CEs) as non-flammable solvents since the beginning of the 20th century. These compounds are known to persist in the environment, exerting adverse effects. They have been considered for a long time as totally resistant to any form of microbial degradation. Advances in this field since the 1980’s showed that several bacteria can couple the reductive dechlorination of CEs to energy conservation in a process called dehalorespiration. Objectives The formation and composition of the microbial habitat, as well as the structure of the bacterial communities (in terms of cell density and species composition), are thought to be related directly to the dehalorespiration activity and to have an influence on the dehalorespiring guild diversity. Methods The bacterial communities present in 104 water samples coming from nine Quaternary European aquifers were analyzed using a standardized T-RFLP analysis. These aquifers showed various sizes and depths and were all contaminated by substantial amounts of PCE. Chemical and contaminant compositions were recorded as well. Results Strong relationships have been observed between the structure of the communities and both geological and environmental data sets, showing an apparent perfect adequacy between the bacterial cells and their specific habitat. In general, each aquifer hosted a specific bacterial community, whose variation was lower than the variation measured among all aquifers. No clear geographical pattern could be found among all aquifers, although temperature was an important variable that contributed apparently to the structuring of the communities. Surprisingly, statistical analysis showed that CEs did not play an important role in the structuring of the communities, even though a partial or total reductive dehalogenation process could be observed in all aquifers. T-RFs corresponding putatively to already known guild members could be detected in almost all water samples, but they contributed marginally (10 ± 7.4 %, in 85 water samples) to the community profiles. Furthermore, these T-RFs showed a remarkably low fidelity index and were consequentlnever linked specifically to any of the observed aquifers