Cessation of water flow from spring snow melt in Dry Creek, a mid-elevation intermit- tent stream in northern Arizona, produces three types of aquatic habitats: main channel ponds, seepage pools, and isolated pools. Differences among the habitat types permit coexistence between a predaceous dytiscid larva and a mosquito larva along this streamcourse. Experiments showed that mosquito populations in habitats that lacked the predator later became extinct when the predator was introduced. The predator is mobile and can move to all three types of habitats, two of which are rich in mosquito prey. The dytiscid did not normally exist in these two habitats during the day. Both types of pools were smaller and had less silt and debris than did side pools along the main channel. Lack of substantial substrata may be a factor contributing to the absence of the predator, as the predator also is cannibalistic and has a tendency to hide within the substrata after encountering another individual of the same species. Information also is presented relating numbers of species to the types and sizes of pools censused in the study. The smaller pools appeared to exclude predators, making these pools more favorable habitats for the two common prey, Culiseta inorata and tadpoles of Hyla arenicolor. Since the formalization of predator-prey relations by Lotka (1925), their inherently oscillatory nature has been universally recognized. But along with this recognition has developed an enigma-inherent forces in a predator-prey system frequently result in the extinction of the system itself. Numerous mecha- nisms have been proposed that are capable of accounting for the persistence of predator-prey pairs under natural conditions. The effect of migration has been described theoretically (Segal and Jackson, 1972), and the presence of some sort of refuge accessible to the prey but not to the predator has been shown to stabilize predator-prey systems under laboratory conditions (Huffaker, 1958). The com- bined effect of habitat assessibility and migration has not yet been demonstrated in nature. We provide evidence that migration and habitat suitability seem to be involved in the maintenance of a predator-prey system in a natural situation. We also present data relating number of taxa found in different sizes and types of pools.