A recent model, designed with stream systems in mind, suggested that prey exchange (movement of prey among patches) tends to reduce predator impacts on prey density; that is, rapid prey immigration into a patch with predators can swamp local effects of predators on prey density. The previous model, however, included two assumptions that influence the model's qualitative predictions. First, it assumed that the system's focal patches are surrounded by an environment that has no predators and a constant prey density. More importantly, it assumed that prey do not alter their per capita emigration rates in response to the presence of predators. We extended the earlier model by relaxing these assumptions. Specifically, we: (1) addressed predator impacts in patches surrounded by background environments that do not have predators, (2) allowed the background environment to have a constant or decreasing prey density (i.e., we examined situations in which predators deplete prey), and (3) accounted for the fact that prey per capita emigration rates are often altered by the presence of predators. Our most interesting results concerned the profound effects of prey emigration behavior on the relationship between prey exchange and predator impact. If prey do not alter their emigration rates in response to predators, then, as predicted by the earlier model, high prey exchange should result in very low predator impact. If, however, prey disperse out of patches in response to the presence of predators (i.e., if prey per capita emigration rates are higher out of patches with predators than out of predator—free patches), then even very high prey exchange rates cannot swamp predator impact; instead, prey emigration adds to predator impact. Thus, depending on prey emigration behavior, increased overall prey exchange can either decrease or increase predator impact. Predators can also suppress prey emigration (e.g., if prey hide in refuges so effectively that they disperse at low rates from predator patches). In that case, high prey exchange rates tend to result in “negative predator impacts” (i.e., higher prey density in patches with predators). The details of the above relationships are influenced by whether or not the background environment has predators. If the background has predators, but predators do not deplete prey (e.g., if predation is offset by recruitment), then predator impact depends only on the ratio of per capita emigration rates out of predator and predator—free patches; that is, attack rates do not influence impact. In contrast, if predators can deplete prey, then attack rates influence predator impact. In that situation, if attack rates are highly relative to prey emigration rates out of predator—free patches, then predator impact steadily increases over time. A literature review suggested that prey alterations in per capita emigration rates in response to the presence of predators can potentially explain some surprising natural phenonomena, including the existence of “negative predator impacts,” and the apparent tendency for invertebrate predators (primarily, stoneflies) to have stronger impacts, and the apparent tendency for vertebrate predators (primarily, fish). Finally, we discussed possible adaptive links between prey escape success, refuge use, dispersal behavior, and predator impacts. This discussion raised a “paradox of danger” that due to their effects on prey exchange, more dangerous predators might often have unexpectedly weak effects on local prey density. In this context, we suggest a framework for studying relationships between prey behavior, prey exchange, and predator impacts.