Habitat isolation is well known to alter patterns of species' abundance, richness, and the ratios of predators : prey. Less clear, however, is how isolation alters interactions within food webs. Here, we present the results from an experiment performed in artificial ponds (mesocosms) manipulating habitat isolation crossed with a predator reduction treatment to disentangle how isolation mediates the top-down effect of predators. The strength of the trophic cascade, from predators, through herbivores, to producers, was considerably stronger in connected than in isolated habitats. We further found that the overall richness of both predator and herbivore species declined strongly with isolation. Experimental predator reductions suggest that the mechanism underlying the herbivore response was likely mediated by a keystone predator effect; when predators were reduced, herbivore richness was lower, and there was no discernable effect of isolation on herbivore richness. Finally, we found that the composition of predators in more isolated habitats consisted of species that were smaller and likely less effective predators than species that persisted in less isolated habitats. In all, our experiment showed that habitat isolation can alter the structure of communities by a combination of direct effects of the species in question, as well as effects mediated through their interactions in the food web.