Considerable effort has been expended to determine if crops genetically engineered to produce Bacillus thuringiensis (Bt) toxins harm non-target arthropods. However, if Bt crops kill target pests and thereby reduce insecticide use, this could benefit some non-target arthropods. We analyzed data from 21 commercial cotton fields in Arizona to test the effects of Bt cotton on insecticide use and abundance of two non-target arthropods, the generalist predators Chrysoperla carnea Stephens (Neuroptera: Chrysopidae) and Orius tristicolor (White) (Heteroptera: Anthocoridae). The number of insecticide sprays was more than double for non-Bt cotton compared with Bt cotton that produced Cry1Ac. The abundance of both predators was negatively associated with the number of insecticide sprays, although significantly so for only one of two sampling periods for each species tested. With the effects of insecticides statistically removed, field type (Bt or non-Bt cotton) did not affect the abundance of either predator. Accordingly, without adjusting for the effects of insecticide sprays, the abundance of C. carnea was higher in Bt cotton fields than in non-Bt cotton fields, but significantly so during only one of two sampling periods. The abundance of O. tristicolor did not differ between field types, even without adjusting for effects of insecticide sprays. The results indicate that Bt crops can affect insecticide use, which in turn can affect the relative abundance of non-target arthropods in Bt and non-Bt fields. Thus, environmental impact assessment should incorporate analysis of the effects of transgenic crops on management practices, as well as evaluation of the direct effects of such crops.