The use of pesticides in agricultural soils may affect the soil microbiota. The effect of repeated application of copper sulfate in soil on indigenous populations of rhizobia was assessed in a medium-term field experiment. Copper sulfate was applied over 8 years at two different rates, 12.5 and 50 kg of CuSO4 ha(-1) year(-1), in the field. The concentrations of total copper in soil varied between 14.0 (control plots that did not receive copper sulfate) and 91.0 mg kg(-1) (the most contaminated plots) at the time of sampling, 3 years after the end of the copper treatments. All the other physicochemical parameters were similar among the plots that also shared the same cropping history. The target rhizobia were monospecific populations of Rhizobium leguminosarum bv. viciae nodulating Vicia sativa and communities of rhizobial species nodulating Phaseolus vulgaris. The size of the vetch rhizobial populations was significantly reduced in the soils with the higher Cu content, whereas the size of the Phaseolus rhizobial populations was not significantly affected. However, the number of nodules formed on both vetches and common beans were reduced for the plants grown in the most contaminated soils, suggesting an additional toxic effect of copper on plant physiology. The diversity (Simpson's indices) of rhizobial genotypes, as characterized by polymerase chain reaction restriction fragment length polymorphism of 16S-23S rDNA intergenic spacer (IGS), was not influenced by copper application. Also, the genetic structure of the R. leguminosarum bv. viciae populations was not modified by copper treatments. By contrast, a shift was observed in the composition of the Phaseolus-nodulating communities in relation to soil copper content. The communities were composed of three 16S rDNA haplotypes: one corresponding to the R. leguminosarum (biovar phaseoli) species, the two others forming a new lineage of Phaseolus rhizobia based on 16S rDNA sequence analysis. The reduced frequency of the R. leguminosarum species in the Phaseolus-nodulating communities from the copper-treated soils was linked to its higher sensitivity to copper as compared to the higher tolerance of isolates belonging to the other rhizobial lineage. The new lineage was functionally efficient for symbiotic nitrogen fixation with P. vulgaris. Our results suggest that functional redundancy among species exhibiting variability for copper tolerance preserved the size of Phaseolus-nodulating communities. In contrast, the abundance of the vetch-nodulating rhizobia, which was a monospecific functional group mainly constituted by copper-sensitive genotypes, was adversely affected by repeated application of copper sulfate.