Trophic cascades are important drivers of plant and animal abundances in aquatic and aboveground systems, but in soils trophic cascades have been thought to be of limited importance due to omnivory and other factors. Here we use a meta-analysis of 215 studies with 1526 experiments that measured plant growth responses to additions or removals of soil organisms to test how different soil trophic levels affect plant growth. Consistent with the trophic cascade hypothesis, we found that herbivores and plant pathogens (henceforth pests) decreased plant growth and that predators of pests increased plant growth. The magnitude of this trophic cascade was similar to that reported for aboveground systems. In contrast, we did not find evidence for trophic cascades in decomposer-and symbiont-based (henceforth mutualist) food chains. In these food chains, mutualists increased plant growth and predators of mutualists also increased plant growth, presumably by increasing nutrient cycling rates. Therefore, mutualists, predators of mutualists and predators of pests all increased plant growth. Further, experiments that added multiple organisms from different trophic levels also increased plant growth. As a result, across the dataset, soil organisms increased plant growth 29% and non-pest soil organisms increased plant growth 46%. Omnivory has traditionally been thought to confound soil trophic dynamics, but here we suggest that omnivory allows for a simplified perspective of soil food webs – one in which most soil organisms increase plant growth by preying on pests or increasing nutrient cycling rates. An implication of this perspective is that processes that decrease soil organism abundance (e.g. soil tillage) are likely to decrease aboveground productivity. There is a growing appreciation for the fact that changing soil organism abundance and community composition can change plant productivity (Bever et al. 2010, Eisenhauer et al. 2012, Kulmatiski et al. 2012, Van der Putten et al. 2013). However, due to the difficulties inherent in working in dense, dark and diverse soils, most research on plant–soil interactions either treats soils as a 'black-box' (i.e. plant– soil feedback research) or focuses on the effects of specific soil pathogens or symbionts (e.g. plant disease research). Less is known about how ecological interactions in soils affect plant growth (Wardle 2006, Fierer et al. 2009, Strong and Frank 2010, Schneider et al. 2012). As a result, the consequences of intended (e.g. pesticide use) and unintended (e.g. climate) changes to soil communities are poorly understood (Lang et al. 2014). In aquatic and aboveground terrestrial systems, trophic cascades help explain how biomass is distributed among plants and animals (Hairston et al. 1960, Paine 1980, Borer et al. 2005, Schmitz et al. 2006, Strickland et al. 2013). Trophic cascades are expected to be more important in sys-tems with 1) homogeneous habitats, 2) fast prey population