The competitiveness of Bradyrhizobium japonicum strains G49 and SMGS1 was first studied in the greenhouse in sterilized sand, with or without added soil. Strain SMGS1 was more competitive than strain G49 with soybean (Glycine max L.) cultivar Labrador but the two strains showed equivalent competitiveness with cultivar Kingsoy. When soil was added, nodule occupancy of strain G49 was only 22% with this cultivar. In field experiments, conducted over 2 years in soils already containing strain G49 (1.5 × 103 to 4.0 × 104 cells/g of soil), nodule occupancy of inoculated strain SMGS1 ranged from 20 to 90%. Nodule occupancy was 3–22% higher when inoculation was done by peat seed coating or with liquid inoculation in the row than with peat-coated clay microgranulars. Nodule occupancy was also dependent on the physiological state of the inoculated cells. When an inoculum stored at 28 °C for 1 year was used at the same viable cell rate, nodule occupancy of strain SMGS1 was 4–20% lower than with a recently made inoculum. Pot experiments with soil from field experiments carried out in the 1st year showed that the inoculated strain continued forming nodules without further inoculation, with a recovery rate equivalent to that of field experiment in the previous year.Key words: Bradyrhizobium japonicum, interstrain competition, inoculation technology, ELISA, field trials.