Spatial distributions of Russian wheat aphid,Diuraphis noxia(Mordvilko), populations on preferred and nonpreferred hosts, as predicted by a spatial simulation model, were evaluated using geostatistics. The purpose was to assess effects of different hypothesized plant-to-plant movement behaviors on population spatial structure and to determine which behavior produced spatial structures (associations) most similar to those observed experimentally. The simulation model predicted location, chronological age, reproduction, movement, and mortality of individual aphids, at 1-d time intervals, within a 25-by-25 plant grid of preferred (wheat) or nonpreferred (oat) hosts. Three aphid movement behaviors were simulated: random, directional, and aggregative. Aggregativc movement produced spatial dynamics closest to the observed experimental data of the three simulated behaviors. Directional movement was the most sensitive to stochasticity in the model. The Taylor power law was tested and found to be not useful for assessing the spatial simulation. We question the utility of this method for assessing spatial relationships and patterns; however, indices of aggregation are useful for the design of sampling protocol. This study demonstrates the utility of geostatistics for assessing the ability of a spatial simulation to model dynamics of a spatial phenomenon.