Selected Bradyrhizobium japonicum strains inoculated on soybean seeds often fail to occupy a significant proportion of nodules when a competitor rhizobial population is established in the soil. This competition problem could result from a genetic/ physiological advantage of the adapted soil population over the introduced inoculant or from a positional advantage, as the soil population already occupies the soil profile where the roots will penetrate, whereas the inoculant remains concentrated around the seeds. Here, we have assessed the contribution of these factors with a laboratory model in which a rhizobial population is established in sterile vermiculite. We observed that the wild-type strain B. japonicum LP 3004 was able to grow in pots with N-free plant nutrient solution-watered vermiculite for six or seven generations with a duplication rate of at least 0.7 day1. In addition, the rhizobial population persisted for 3 months with 106–107 colony-forming units ml1 of the vermiculite-retained solution. N-starved, young rhizobial cultures are more efficient in performing several steps along their early association with soybean roots. However, N starvation during growth of rhizobia used for seed inoculation did not enhance their competitiveness against a 1 month vermiculite-established rhizobial population, which occupied more than 72% of the nodules. When a similarly established rhizobial population was recovered from the vermiculite and homogeneously suspended in plant nutrient solution, these cells were significantly less competitive (29% of nodules occupied) than rhizobia obtained from a fresh, logarithmic culture in a N-poor minimal medium, thus indicating that cell position rather than intrinsic competitiveness was the determinant for nodule occupation.