The literature on herbivorous insects in heterogeneous habitats has addressed insect population responses to patch size, distance from source populations, habitat edges, and variation in host stem density. Studies typically conclude that insect colonists respond positively to the area of host plant patches, but there is little consensus on how insects respond to variation in host density at the patch level. Although observed variation likely results partially from differences in study species and focal habitats, deviations from expectations also may reflect the importance of unmeasured habitat parameters. In this study of colonization by cicada Magicicada cassinii (Fisher) in an experimentally fragmented old field, we simultaneously examined landscape variation in patch size, distance from the sources of colonization, and local host plant stem density (at a within-patch scale) and also considered edge effects. Per stem colonization was positively related to proximity to the population source and habitat patch size but negatively related to local host stem density. The effect of edge was nonsignificant. When coarser scale estimates of local stem density (calculated by averaging the figures for all quadrats within patches) were used in analyses, fewer significant main effects were found and sometimes interactions occurred. Our study highlights the importance of including all potential explanatory variables in analyses, with what we feel is a novel observation that the explicit consideration of fine-scale, within-patch variation in local stem density can be important to the interpretation of insect dispersal and colonization.