Ecologists have long known that multiple predator species can interact with each other and thereby either strengthen or weaken overall prey regulation. With few exceptions our understanding of such ‘multiple predator effects’ (MPEs) is based on experimental combinations of predators at a single relative density (usually 1:1). Because MPEs depend on interspecific interactions between predators, we hypothesized they would vary, potentially non-linearly, with predator species relative abundance. We tested this hypothesis in a southeastern US salt marsh by manipulating two species of predatory crab to generate a continuous relative abundance gradient. After four months, we evaluated the density of two shared prey species (snails and fiddler crabs) across this gradient, before explicitly testing for: 1) the presence of overall MPEs on the densities of these prey; 2) whether (and how) potential MPEs varied as a function of relative abundance; and 3) how indicators of predator-predator interactions (survivorship and limbs lost in contests) were affected by relative abundance. The final density of both prey species varied with relative abundance, but the sign of these effects switched depending on prey identity. The results failed to support an overall MPE on snail density, but final fiddler crab density was higher than expected (i.e., risk reduction, or an overall negative MPE on fiddler crab suppression). Counter to our prediction, this MPE did not vary as a function of relative abundance. Predator survivorship and limb loss indicated asymmetrical negative interactions that strongly impacted the predator species most effective at suppressing fiddler crabs, suggesting an explanation for the negative MPE observed for this prey species. Our findings suggest that MPEs are not always sensitive to species relative abundance, but given that shifts in predator relative abundance are frequently observed in nature, future studies should incorporate this aspect of biodiversity change into their designs wherever possible.