A 2-year commercial-scale study was conducted to qualitatively evaluate the effect of different insecticide treatment regimes on the predator complex attacking Bemisia tabaci (Gennadius) in cotton. In 1996 three insecticide regimes were compared: a rotation of conventional broadspectrum insecticides or one of two different regimes based on the initial use of the insect growth regulators (IGRs), buprofezin and pyriproxyfen. In 1997 the same three regimes plus an untreated control were compared; split-plots were sprayed once for Lygus hesperus Knight control using a broad-spectrum insecticide. Relative feeding activity for each predator species was compared between treatment regimes by analyzing the gut contents of predators for the presence of whitefly remains using a whitefly-specific enzyme-linked immunosorbent assay (ELISA). The ELISA results were combined with predator density data to obtain a qualitative pesticide impact index for each predator group. In total, we analyzed the gut contents of 32 262 field-collected predators, representing nine different taxa. Of these, Pseudatomocelis seriatus (Reuter), Spanagonicus albofasciatus (Reuter), and spiders consisting primarily of Misumenops celer (Hentz) are shown to be whitefly predators for the first time. Predator populations were usually reduced in plots that received applications of broad-spectrum insecticides for B. tabaci and L. hesperus control, but there were few treatment differences in the proportions of predators containing whitefly remains in their guts. However, the feeding activity of certain predator species in fields sprayed with broad spectrum insecticides was significantly reduced compared with those in IGR-based and control treatments. Overall, insecticide regimes using IGRs were less lethal to the whitefly predator complex than regimes consisting of only conventional, broadspectrum insecticides, but differences in predator feeding activity on whitefly between the various insecticide treatment regimes were minimal.