This is the report of the first open field release of genetically modified microorganisms (GMMs) in Italy. It covers ten years of monitoring, and follows in-field GMM dynamics from strain release to disappearance below detection limits, as well as assessment of impact on resident microorganisms. The bacteria released belong to the nitrogen fixing legume endosymbiont Rhizobium leguminosarum bv. viciae, and were engineered with non-agronomically-proficient traits, in order to assess their behavior and fate without GMM-specific positive feedback from the plant. A DNA cassette containing mercury resistance and ß-galactosidase genes was introduced in either plasmid-borne or chromosomally integrated versions, in order to test the resulting strain stability. A synthetic promoter was used to drive the lacZ gene, conferring high catabolic activity to the GMM. Two different wild-type Rhizobium backgrounds were tested, comparing a non-indigenous vs. an indigenous, highly competitive strain. The latter had much greater persistence, since it was able to survive and establish at technically detectable levels for over four years after release. Selection factors, such as reiterated presence of the plant host, or lactose substrate supply, enhanced long-term survival to different extents. The lactose treatment showed that even a single trophic supplementation can surpass the benefits of symbiotic interaction for a period of several years. Concerning impact, the GMMs did not alter substantially the other soil community general microbiota. However, there were some significant differences in microbiota as a consequence of the Rhizobium inoculation. This effect was observed with either the WT or GMM, and was more evident in the release of the indigenous Rhizobium. Moreover, as the indigenous GMM had its parental, dominant wild-type in the same soil, it was possible to evaluate to what extent the GMM version could result in parent displacement (“self-impact”), and how much the two rhizobia would additively contribute to nodulation.