We assessed the effects of EDTA and selected plant growth-promoting rhizobacteria (PGPR) on the phytoremediation of soils and sediments historically contaminated by Cr, Ni, and Cu. A total of 42 bacterial strains resistant to these heavy metals (HMs) were isolated and screened for PGP traits and metal bioaccumulation, and two Enterobacter spp. strains were finally selected. Phytoremediation pot experiments of 2 months duration were carried out with hemp (Cannabis sativa L.) and giant reed (Arundo donax L.) grown on soils and sediments respectively, comparing in both cases the effects of bioaugmentation with a single PGPR and EDTA addition on plant and root growth, plant HM uptake, HM leaching, as well as the changes that occurred in soil microbial communities (structure, biomass, and activity). Good removal percentages on a dry mass basis of Cr (0.4%), Ni (0.6%), and Cu (0.9%) were observed in giant reed while negligible values (<100‰) in hemp. In giant reed, HMs accumulated differentially in plant (rhizomes > > roots > leaves > stems) with largest quantities in rhizomes (Cr 0.6, Ni 3.7, and Cu 2.2 g plant–1). EDTA increased Ni and Cu translocation to aerial parts in both crops, despite that in sediments high HM concentrations in leachates were measured. PGPR did not impact fine root diameter distribution of both crops compared with control while EDTA negatively affected root diameter class length (DCL) distribution. Under HM contamination, giant reed roots become shorter (from 5.2 to 2.3 mm cm–3) while hemp roots become shorter and thickened from 0.13 to 0.26 mm. A consistent indirect effect of HM levels on the soil microbiome (diversity and activity) mediated by plant response (root DCL distribution) was observed. Multivariate analysis of bacterial diversity and activity revealed not only significant effects of plant and soil type (rhizosphere vs. bulk) but also a clear and similar differentiation of communities between control, EDTA, and PGPR treatments. We propose root DCL distribution as a key plant trait to understand detrimental effect of HMs on microbial communities. Positive evidence of the soil-microbe-plant interactions occurring when bioaugmentation with PGPR is associated with deep-rooting perennial crops makes this combination preferable over the one with chelating agents. Such knowledge might help to yield better bioaugmented bioremediation results in contaminated sites.