The environmental impact of genetically modified (GM) plants in experimental fields has been examined in several ways, in particular with respect to the dynamics of specific nontarget organisms. The approach of sampling for biodiversity in agroecosystems to compare complex patterns could also be useful in studying potential disruptions caused by GM crops. In this study, we set up replicated field plots of Bt-expressing eggplants and near isogenic untransformed eggplants as a control. We monitored the presence and abundance of herbivore and predator arthropods in weekly visual samplings of the plant canopy for three growing seasons (2001–2003). Insect species were pooled in organismal taxonomic units (OTUs); three multivariate methods were used to compare species assemblage as an estimate of insect biodiversity. This multistep statistical approach proved to be efficient in recognizing association patterns, as evidenced by the data for the target species Leptinotarsa decemlineata Say (Coleoptera: Chrysomelidae) clearly showing a significant association with the control plots. All the analyses indicate a comparable species assemblage between transgenic and near isogenic eggplant areas. Our results suggest that some taxa may warrant more specific study. For example, Alticinae beetles (Coleoptera: Chrysomelidae) were alternatively more abundant in either of the two treatments, and their overall abundance was significantly higher on transgenic eggplants. In light of these results and because of their taxonomic proximity to the target species, these herbivores may represent an important nontarget group to be further studied. Moreover, some sap feeders (e.g., Homoptera: Cicadellidae) were more abundant on Bt-expressing plants in some samples in all 3 yr.