Transgenic (Bt) cotton dominates Australian cotton production systems. It is grown to control feeding damage by lepidopteran pests such as Helicoverpa armigera. The possibility that these moths might become resistant to Bt remains a threat. Consequently, refuge crops (with no Bt) must be grown with Bt cotton to produce large numbers of Bt-susceptible moths to reduce the risk of resistance developing. A key assumption of the refuge strategy, that moths from different host plant origins mate at random, remains untested. During the period of the study reported here, refuge crops included pigeon pea, conventional cotton (C3 plants), sorghum or maize (C4 plants). To identify the relative contributions made by these (and perhaps other) C3 and C4 plants to populations of H. armigera in cotton landscapes, we measured stable carbon isotopes (δ13C) within individual moths captured in the field. Overall, 53% of the moths were of C4 origin. In addition, we demonstrated, by comparing the stable isotope signatures of mating pairs of moths, that mating is indeed random amongst moths of different plant origins (i.e. C3 and C4). Stable nitrogen isotope signatures (δ15N) were recorded to further discriminate amongst host plant origins (e.g. legumes from non-legumes), but such measurements proved generally unsuitable. Since 2010, maize and sorghum are no longer used as dedicated refuges in Australia. However, these plants remain very common crops in cotton production regions, so their roles as ‘unstructured’ refuges seem likely to be significant.