Ecologists have long debated the role of predation in mediating the coexistence of prey species. Theory has mainly taken a bitrophic perspective that excludes the effects of inducible defenses at different trophic levels. However, inducible defenses could either limit or enhance the effects of predation on coexistence, by means of effects on bottom-up control and population stability. Our aim was to investigate how inducible defenses at different trophic levels affect the possibilities for predator-mediated coexistence, as opposed to competitive exclusion, in replicated experimental plankton communities. In particular, we analyzed how the presence or absence of inducible defenses in algal basal prey affected the outcome of competition between an inducible defended and an undefended herbivore, in the presence or absence of a carnivore. We found the undefended herbivore to be a superior competitor in the absence of predation. This outcome was reversed in the presence of a shared carnivore: populations of the undefended herbivore then strongly declined. The extent of this population decline differed between food webs based on undefended as opposed to inducible defended algal prey. In the former the undefended herbivore became undetectable for most of the duration of the experiment. In the latter the undefended herbivore also crashed to low densities, but it could still be detected during most of the experiment. In food webs based on inducible defended algae, the carnivore failed to reach high densities and exerted weaker top–down control on the two competing herbivores. We conclude that the inducible defense in one of our two competing herbivores allowed the outcome of competition to be reversed when a shared carnivore was added. Inducible defenses in algae did not change this outcome, but they significantly delayed extinction of the undefended herbivore. Predation itself did not promote coexistence in these experimental plankton communities.