Transgenic cotton varieties (Bollgard II) expressing two proteins (Cry1Ac and Cry2Ab) from Bacillus thuringiensis (Bt) have been widely adopted in Australia to control larvae of Helicoverpa. A triple-stacked Bt-transgenic cotton producing Cry1Ac, Cry2Ab, and Vip3A proteins (Genuity Bollgard III) is being developed to reduce the chance that Helicoverpa will develop resistance to the Bt proteins. Before its introduction, nontarget effects on the agro-ecosystem need to be evaluated under field conditions. By using beatsheet and suction sampling methods, we compared the invertebrate communities of unsprayed non-Bt-cotton, Bollgard II, and Bollgard III in five experiments across three sites in Australia. We found significant differences between invertebrate communities of non-Bt and Bt (Bollgard II and Bollgard III) cotton only in experiments where lepidopteran larval abundance was high. In beatsheet samples where lepidopterans were absent (Bt crops), organisms associated with flowers and bolls in Bt-cotton were more abundant. In suction samples, where Lepidoptera were present (i.e., in non-Bt-cotton), organisms associated with damaged plant tissue and frass were more common. Hence in our study, Bt- and non-Bt-cotton communities only differed when sufficient lepidopteran larvae were present to exert both direct and indirect effects on species assemblages. There was no overall significant difference between Bollgard II and III communities, despite the addition of the Vip gene in Bollgard III. Consequently, the use of Bollgard III in Australian cotton provides additional protection against the development of resistance by Helicoverpa to Bt toxins, while having no additional effect on cotton invertebrate communities.