We conducted manipulative field experiments in artificial ponds to quantify the predatory impact of larvae of a migratory dragonfly (Tramea lacerata) on a common resident dragonfly species (Erythemis simplicicollis), and on damselflies as shared prey of the two dragonflies. We found that the combined predatory effects of these two dragonflies on damselflies were not additive. To determine the underlying cause of non—additive predation rates in the field, we conducted a second experiment in laboratory aquaria to isolate the impact of each predator on the consumption rates of the other. Dragonfly consumption rates of damselflies in single—predator treatments were compared to those in the presence of heterospecifics or conspecifics with their menta (mouthparts) surgically modified so that they could not capture prey. In the laboratory experiment, de—mented Tramea reduced the consumption rates of Erythemis to less than half of that observed when Erythemis foraged alone. Erythemis numbers were also reduced by Tramea predation. Erythemis had neither effect on Tramea. Both of the negative effects of Tramea on Erythemis will have indirect positive effects on damselflies. The "behavioral" component (reduced Erythemis foraging rate) should be more important than the "trophic link" (reduced Erythemis numbers) indirect effect. Together these indirect positive effects will allay, but not completely compensate for, the direct negative effects of Tramea predation on damselflies. These results illustrate how an asymmetric potential for intraguild predation can lead to asymmetries in interference competition and to non—additive effects on prey mortality. The addition of removal of predators that interact in this manner to or from communities should have only a small net effect on prey because of compensating direct and indirect effects. This may explain why predator manipulations have often had unpredictable or undetectable effects on freshwater benthic communities.