A range of soybean commercial inoculants were tested under greenhouse and field conditions in different locations in Kenya. One rhizobial (Legumefix) and one mycorrhizal (Rhizatech) inoculants were selected according to the promising results obtained. The efficiency of these 2 products (single or combined inoculation) was assessed in 150 farms in 3 mandate areas in Kenya presenting different soil characteristics and environmental conditions. Both biomass and grain yields were measured and nodules occupancy of the rhizobial strain was assessed by ELISA using specific monoclonal antibodies. Application of the rhizobial inoculant, alone or in combination with the mycorrhizal product significantly increased the yields in all mandate areas (about 75% of the farms). However, the co-inoculation didn't perform significantly better than the rhizobial product alone. Controls plots gave poor results, as well as the mycorrhizal product, though that was not expected given the good performances observed under greenhouse conditions. Nodule occupancy analysis showed that a high number of nodules occupied by the inoculated strain did not obviously lead to an increase of soybean production. Soil factors (pH, P, C, N…) seemed to affect the inoculant efficiency whether the strain is occupying the nodules or not. Our statistic analysis showed that soil pH significantly affected nodulation and yield, though the effect was variable depending on the region. P content also positively correlated with nodulation and yields. This study provided relevant information on field efficiency of rhizobial inoculants on soybean yield in Kenya and showed that the competitiveness of rhizobial strains might not be the main factor explaining the effect (or lack of) of legumes inoculation in the field. (Texte intégral)