Field studies were conducted from 2000 through 2002 to determine whether the abundance of predaceous arthropods was negatively affected by Bollgard and Bollgard II transgenic cotton, Gossypium hirsutum (L.), that expresses one or two insecticidal proteins of Bacillus thuringiensis Berliner (Bt) for control of lepidopteran pests. During 2000, populations of predators and lepidopteran pests were monitored in small plots of DP50 (non-Bt), DP50B (Bollgard, expresses Cry1Ac), and DP50BII (Bollgard II, expresses Cry1Ac and Cry2Ab) under untreated conditions. In subsequent studies, plot sizes were increased and insect populations were monitored under conditions in which predators were either allowed to develop normally (untreated) or were disrupted with insecticides in early season. Populations of predators (Geocoris spp., Orius insidiosus [Say], Nabis spp., Solenopsis invicta Buren, spiders, coccinellids, and lacewings [Chrysopidae and Hemerobiidae]) were consistently as high or higher in Bollgard and Bollgard II compared with non-Bt. Under untreated and disrupted conditions, season-long populations of Helicoverpa zea (Boddie) were consistently but not significantly lower in DP50BII compared with DP50B. In this study and in previous studies, DP50BII provided excellent control of H. zea despite predator disruption. However, under high population pressure in 2002, H. zea numbers exceeded economic thresholds when predators were disrupted (five larvae >2/3 cm in length per 100 plants on 26 July) and would have required insecticidal intervention. Compared with DP50B, DP50BII provided significantly better control of Pseudoplusia includens (Walker) and Spodoptera frugiperda (J.E. Smith) under both disrupted and untreated conditions. Until experience is gained in wide scale commercial plantings, Bollgard II varieties should be monitored closely for lepidopteran pests that might survive the toxins, particularly where populations of predaceous arthropods have been disrupted.