1. Interactions between herbivores and their predators are shaped, in part, by plant phenotype. Consequently, ubiquitous symbionts of plants below-ground, such as arbuscular mycorrhizal fungi (AMF), may influence interactions above-ground between predators and their prey by altering plant phenotype. However, the ecological relevance of below-ground organisms on predator-prey interactions under field conditions remains unclear.2. We assessed how AMF influence herbivore-predator interactions through a field experiment.3. We planted two milkweed species (Asclepias curassavica and Asclepias incarnata) provided with different amounts of AMF inoculum (zero, medium, and high) in a randomized block design. We added aphids to plants and reduced predator pressure weekly for 5 weeks to evaluate effects of AMF on predator recruitment.We then allowed herbivore-predator interactions to re-establish naturally for the remainder of the season to examine whether AMF-mediated variation in predator recruitment influenced the suppression of aphid populations. 4. Arbuscular mycorrhizal fungi availability in soils mediated interactions between predaceous aphid midge flies Aphidoletes aphidimyza and their aphid prey Aphis nerii, but the effects were plant species-specific. On A. curassavica, by mid-season, midges were recruited most strongly on plants under medium AMF availability and least on plants under high AMF availability. In contrast, each midge killed fewer aphids with increasing aphid density on medium AMF plants, but killed more aphids with increasing aphid density on high AMF plants. In combination, aphid mortality rates imposed by midges were greatest on medium AMF plants, followed by high and zero AMF plants. By comparison, on A. incarnata, the recruitment of midges was strongest on high AMF plants and weakest on medium AMF plants. AMF had no effect on the number of aphids killed per midge, relative to aphid density, so mortality rates of aphids imposed by midges mirrored recruitment. Rates of decline in aphid populations following predator recolonization were associated with midge densities, as well as lacewing and syrphid densities, which were unaffected by AMF availability. Therefore, the effects of AMF on aphid population decline were not a simple function of AMF-midge interactions. 5. Our findings demonstrate that the availability of AMF in soils has pervasive, but complex, effects on predator-herbivore dynamics in the field.