A long-term investigation of rangeland grasshopper assemblages in southeastern Wyoming has been undertaken to determine if: 1) population dynamics are eruptive or gradient, and 2) suppression of localized hot-spots (i.e., an infestion of economically damaging rangeland species at a density of ≥ 9.6 grasshoppers/ m2 originating in an area of 20 to 4,000 ha) will prevent large-scale outbreaks. Data from 1990-1994 (a regional interoutbreak period) suggest elements of both eruptive and gradient dynamics. The eruptive model is supported by spatial qualities of hot-spots (they are discrete areas averaging 820 ha), dynamics of untreated hot-spots (they can persist and expand), dynamics of treated hot-spots (they do not recover in the treated or adjacent areas in subsequent years), and pest population ecology (at least one species apparently has eruptive reproductive potential). The gradient model is supported by dynamics of outbreaks (infestations >7,000 ha were found without a preceding hot-spot), dynamics of hot-spots (most collapse without treatment), and properties of infested lands (hot-spots have consistent topographic and soil properties). The eruptive-gradient dichotomy may fail to adequately describe grasshopper outbreak dynamics because the models were developed from simple systems that are qualitatively different than rangeland grasshoppers. Based on existing data, we recommend that incipient infestations should be treated only if they exceed 250 ha and they persist for at least 2 yr. Preventative benefits of small-scale treatments are uncertain at this time, although preservation and recolonization of beneficial arthropods are enhanced with this approach.