The populations of two coexisting species of Dipodomys (Heteromyidae, Rodentia) were manipulated on 10, large, unenclosed, trapping grids. These manipulations revealed that, although many kangaroo rats are established residents in an area, a large number are transient individuals who quickly occupy vacated habitats. On plots from which residents had been removed, transients settled at rates of up to 5% of carrying capacity per day. These immigrants were invariably of the same species that was removed, indicating a strong element of intraspecific competition with little or no evidence of competition between the species.Trapping records suggest that these species avoid competition through habitat selection. Dipodomys ordii prefer a grassier habitat, and D. merriami a habitat dominated by creosote bush. Apparent overlap in their utilization of habitats, based on sites of capture, predicts competition coefficients to be higher than those permitted by the theory of limiting similarity and much higher than those actually shown by the perturbation experiments.This study demonstrates the dangers of estimating alpha without experimentation. This is especially true in cases where habitat selection may be important, since organisms may travel in habitats without collecting resources therein. Our results are discussed in light of a theory which examines the optimal (rather than tolerable) amount of overlap in habitat utilization between two potential competitors in a mixed habitat. This theory predicts that the pressure of natural selection should eliminate the interspecific competition entirely.However, the conclusion that the interspecific competitive alpha is zero does not lead to the conclusion that interspecific competition is unimportant in the system. Instead, if our interpretation is correct, such competition has molded the system, and were there not a continual threat of interspecific competition, the habitat specializations would soon disappear.