The contamination of agricultural land and groundwater by heavy metals is essentially linked to human activities. A major problem with heavy metals is that they cannot be biodegraded and therefore reside in the environment for long periods of time if they are not removed. Thus, depending of the kind and depth of contamination, different remediation techniques were developed. One of these methods, called "phytoextraction", uses the ability of so-called hyperaccumulating plants to extract high amounts of heavy metals. Accumulation of heavy metals in the environment is a serious concern for animal and human health. At the microscopic scale, heavy metals may have also deleterious effects on bacteria which are the key-players of the different nutrient turnovers in soils. Consequently, ecosystem functioning can be seriously perturbed and the long-term soil fertility may be threatened. The recent development of molecular biology greatly contributed to the discovery of the microbial diversity and its function in the soil. However, to date only a small number of studies used molecular methods to investigate the impacts of heavy metals on the bacterial community. In this thesis, a pot experiment was conducted under controlled conditions with one hyperaccumulating plant (Thlaspi caerulescens) grown in two different soils, a long-term and an artificially heavy metal-contaminated soil. The impact of heavy metals on the microbial community was then investigated with several molecular methods. Moreover, the decrease of the bioavailable heavy metals concentrations in soil due to plant uptake allowed to study the consequences on bacterial community structure and function. Based on the 16S ribosomal RNA and the corresponding gene (16S rDNA), four clone libraries were constructed to retrieve information on the structure of the microbial community and the potentially active part of the microbial community in the rhizosphere of Thlaspi caerulescens grown during three months in a long-term contaminated soil. The data obtained with the two clone libraries (rRNA and rDNA) from the rhizosphere of Thlaspi caerulescens were compared with the bulk soil data to identify any effect of the plant on the soil microbial community structure. Partial sequence analysis of 282 clones revealed that most of the environmental sequences in both soils affiliated with five major phylogenetic groups, the Actinobacteria, α-Proteobacteria, β-Proteobacteria, Acidobacteria and the Planctomycetales. The taxa dominating the bacterial community structure in the bulk soil also dominated the rhizosphere community, indicating that the plant did not exert a major influence on the overall bacterial diversity. However, all dominant taxa, with the exception of the Actinobacteria, were relatively less represented in the rRNA libraries as compared to the rDNA libraries. On the contrary, sequences belonging to the Actinobacteria dominated both bulk and rhizosphere soil libraries derived from rRNA. Seventy per cent of these clone sequences were related to two subgroups of the Rubrobacteria, which was an indication that this group of bacteria was probably metabolically active in heavy metal-contaminated soils. Fluorescence in situ hybridization (FISH) was used for the in situ detection and quantification of selected bacterial groups previously detected in the clone libraries from the rhizosphere of Thlaspi caerulescens. By applying the most general probe EUB338, only 20% of the total rhizosphere microbial community could be detected. Based on this result, it was difficult to conclude with certitude which were the most dominant bacteria in the rhizosphere. However, despite this low detection rate, it was possible to detect the major groups present in the rhizosphere clone libraries using group-specific oligonucleotide probes. As part of our sequences were affiliated to two emerging bacterial groups, the Acidobacteria and the Rubrobacteria, two new probes were designed, Acido228 (specific for the subgroup 1 of the Acidobacterium division) and Rubro198 (specific for the all Rubrobacteria subclass), for the detection of these microorganisms. These two probes were first checked for their specificity with pure cultures and finally applied in the rhizosphere soil allowing for the first time the detection of these organisms in situ. Finally, the impact of heavy metals and a subsequent one-year phytoextraction with Thlaspi caerulescens on the soil microbial community was investigated in an artificially heavy metal-contaminated soil. All the different molecular and culture-dependent techniques used, denaturing gradient gel electrophoresis (DGGE), community level physiological profile (CLPP) and potential ammonium-oxidation measurement showed that the heavy metal addition induced drastic changes in the bacterial community. Moreover, the analysis of the different bacterial DGGE patterns (Bacteria, β-Proteobacteria and ammonia-oxidising bacteria) obtained during this experiment showed that one-year phytoremediation was not sufficient to recover the initial community present in the non-contaminated soil. However, with the CLPP analysis, it was possible to detect a stimulating effect of the plant on a part of the microbial community in both contaminated and non-contaminated soils. The most obvious result was obtained in the contaminated soil where the number of substrates metabolised increased significantly in the presence of the plant as compared to the unplanted contaminated samples. The measurement of the potential ammonium-oxidation was used as a criterion for soil quality. Although this test showed that the ammonia-oxidising bacteria were significantly stimulated in the planted non-contaminated soil samples, the positive effect of the plant on these bacteria was not sufficient to overcome the inhibition induced by the presence of the heavy metals in the contaminated soil, even after one year phytoremediation. To conclude, molecular methods in combination with culture-dependent techniques have proven in this study to be very useful for the detection of the changes induced by the heavy metals in the structure and the function of the microbial community. Moreover, the molecular techniques contributed to the identification of bacteria which could be potentially used for the bioremediation of contaminated soils thus offering new perspectives of investigation and technology development.