There is increasing evidence that interactions between plants and biotic components of the soil influence plant productivity and plant community composition. Many plant–soil feedback experiments start from inoculating relatively small amounts of natural soil to sterilized bulk soil. These soil inocula may include a variety of size classes of soil biota, each having a different role in the observed soil feedback effects. In order to examine what may be the effect of various size classes of soil biota we compared inoculation with natural field soil sieved through a 1 mm mesh, a soil suspension also sieved through a 1 mm mesh, and a microbial suspension sieved through a 20 μm mesh. We tested these effects for different populations of the same plant species and for different soil origins. Plant biomass was greatest in pots inoculated with the microbial suspension and smallest in pots inoculated with sieved soil, both in the first and second growth phase, and there was no significant population or soil origin effect. Plant-feeding nematodes were almost exclusively found in the sieved soil treatment. We show that processing the soil to obtain a microbial suspension reduces the strength of the soil effect in both the first and second growth phase. We also show that the results obtained with inoculating sieved soil and with a soil suspension are not comparable. In conclusion, when designing plant–soil feedback experiments, it is crucial to consider that soil inoculum preparation can strongly influence the observed soil effect.