In soils with a resident population of symbiotically effective rhizobia in sufficient number that legume inoculation is not a requirement for successful legume cropping, greater flexibility may be exercised in the use of legume seed dressings that are toxic to rhizobia. Inoculating crops antecedent to legume crops has been suggested as a method for pre-establishing effective inoculant rhizobia in soil. The extent to which this strategy (pre-inoculation) would remove the need for inoculating legume seed (conventional inoculation) was tested for Bradyrhizobium sp. (Lupinus) and Rhizobium leguminosarum bv. viciae with crops of lupin (Lupinus angustifolius) and faba bean (Vicia faba), respectively. In the glasshouse, in pasteurised sand and red kandosol, the numbers of B. sp. (Lupinus) in the rhizospheres of wheat (Triticum aestivum), canola (Brassica campestris) and clover (Trifolium subterraneum) increased 300'10000-fold over a 14-week period, reaching numbers similar to that achieved on L. angustifolius. These increases were not greatly affected by chemical seed dressings commonly applied to the crops: on wheat, Vincit C and Baytan C; on canola and subterranean clover, Lemat. In the sandy soil, the nodulation of lupin following pre-establishment of rhizobia, drying and mixing of soil, was not improved by conventional inoculation; in the red kandosol nodulation was increased only marginally by conventional inoculation. The results with the glasshouse pot trials warranted further investigation in the field. Under field conditions, when B. sp. (Lupinus) was pre-established with wheat, on a red kandosol in south-western New South Wales, the number of these bacteria surviving in the dry soil at the end of the wheat phase was much lower than in the glasshouse study. In the following season, the nodulation of lupin sown without inoculant, and dependent only on pre-established rhizobia, was significantly reduced, as compared with that on conventionally inoculatlupin. An exception occurred where the lupin was dry-sown 3 weeks before rain, in which case nodulation was comparatively poor even with conventional inoculation. Reduced nodulation was generally consistent with initially fewer rhizobia in the lupin rhizosphere. However, the numbers of rhizobia were eventually similar to those found with conventional inoculation. In the third year, in autumn, B. sp. (Lupinus) was abundant in the soil in all treatments and there were no differences in lupin nodulation between treatments. Similarly, introducing R. leguminosarum bv. viciae on wheat, in an acidic red kandosol in south-western New South Wales, failed to provide as much nodulation of faba bean as was achieved with conventional inoculation. The maximal dry matter of the bean crop was also significantly lower with pre-inoculation as compared to conventional inoculation.