Abstract In the managed agricultural ecosystems of southern Australia, if an edaphic environment is not stressful to root-nodule bacteria (hereafter rhizobia), it is likely to become a competitive environment for nodulation (although not always detrimentally so) soon after the introduction of an inoculated legume. We suggest that stressful environments limit rhizobial communities to less than 100 cells g −1 soil at some time during the season. This overview puts forward the hypothesis that in perturbed ecosystems (i.e. those that are intensively managed) such as in the 25 million ha of the southern Australian grain and grazing belts, the rhizobial community is still substantially immature in an evolutionary sense. The rhizobial community is representative of only a few species, primarily those of Mediterranean origin that were accidentally introduced, or have been fostered by legume development programs, or remnants of the populations associated with native legumes. We consider there is little inter-specific competition for substrates because of this relative immaturity, but suggest that intra-specific competition for nodulation is commonplace wherever abiotic stress is absent. We nominate two primary abiotic stresses that are permanently present that have limited rhizobial colonization or legume nodulation for some species in southern Australia and four secondary (temporary) abiotic stresses. We believe that selection of adapted symbioses, or where warranted adapted elite rhizobial strains or legume host genotypes, can overcome these stress factors. We emphasise that where several abiotic stress factors are present they may act synergistically, but that this net effect is still likely to be symbiosis-specific. We acknowledge that genetic transformation in situ is providing new strain variability with which we must contend. We also put forward the suggestion that opportunities exist for the managed introduction of selected genotypes of agricultural legumes that effectively interact with rhizobial communities to achieve optimal N-fixation. In doing so, we give more precise definition to the widely used terms ‘exclusive’, ‘selective’ and ‘promiscuous’ nodulation.