The culture of transgenic Lotus corniculatus plants producing opines, which are bacterial growth substrates, leads to the selection of rhizospheric bacteria able to utilize these substrates. We have investigated the fate of the opine-utilizing community over time under different experimental conditions following elimination of selective pressure exerted by the transgenic plants. These plants were removed from the soil, which was either left unplanted or replanted with wild-type L. corniculatus or wheat plants. The density of opine-utilizing bacteria in the fallow soils remained essentially unchanged throughout the experiment, regardless of the soil of origin (soil planted with wild-type or transgenic plants). When wild-type Lotus plants were used to replace their transgenic counterparts, only the bacterial populations able to utilize the opines were affected. Long-term changes affecting the opine-utilizing bacterial community on Lotus roots was dependent upon the opine studied. The concentration of nopaline utilizers decreased, upon replacement of the transgenic plants, to a level similar to that of normal plants, while the concentration of mannopine utilizers decreased to levels intermediate between transgenic and normal plants. These data indicate that: (i) the opine-utilizing bacterial populations can be controlled in the rhizosphere via plant-exudate engineering; (ii) the interaction between the engineered plants and their root-associated micro-organisms is transgene specific; and (iii) alterations induced by the cultivation of transgenic plants may sometimes be persistent. Furthermore, opine-utilizing bacterial populations can be controlled by crop rotation. Therefore, favouring the growth of a rhizobacterium of agronomic interest via an opine-based strategy appears feasible.