Many natural ecosystems have been degraded because of human activities1,2 and need to be restored so that biodiversity is protected. However, restoration can take decades and restoration activities are often unsuccessful3 because of abiotic constraints (for example, eutrophication, acidification) and unfavourable biotic conditions (for example, competition or adverse soil community composition). A key question is what manageable factors prevent transition from degraded to restored ecosystems and what interventions are required for successful restoration2,4. Experiments have shown that the soil community is an important driver of plant community development5,6,7,8, suggesting that manipulation of the soil community is key to successful restoration of terrestrial ecosystems3,9. Here we examine a large-scale, six-year-old field experiment on ex-arable land and show that application of soil inocula not only promotes ecosystem restoration, but that different origins of soil inocula can steer the plant community development towards different target communities, varying from grassland to heathland vegetation. The impact of soil inoculation on plant and soil community composition was most pronounced when the topsoil layer was removed, whereas effects were less strong, but still significant, when the soil inocula were introduced into intact topsoil. Therefore, soil inoculation is a powerful tool to both restore disturbed terrestrial ecosystems and steer plant community development.