Regulators are acutely aware of the need for meaningful risk assessments to support decisions on the safety of GM crops to non-target invertebrates in determining their suitability for field release. We describe a process for developing appropriate, testable risk hypotheses for invertebrates in agroecosystems that might be exposed to plants developed by GM and future novel technologies. An existing model (PRONTI) generates a ranked list of invertebrate species for biosafety testing by accessing a database of biological, ecological and food web information about species which occur in cropping environments and their potential interactions with a particular stressor (Eco Invertebase). Our objective in this contribution is to explore and further utilise these resources to assist in the process of problem formulation by identifying potentially significant effects of the stressor on the invertebrate community and the ecosystem services they provide. We propose that for high ranking species, a conceptual food web using information in Eco Invertebase is constructed, and using an accepted regulatory risk analysis framework, the likelihood of risk, and magnitude of impact for each link in the food web is evaluated. Using as filters only those risks evaluated as likely to extremely likely, and the magnitude of an effect being considered as moderate to massive, the most significant potential effects can be identified. A stepwise approach is suggested to develop a sequence of appropriate tests. The GM ryegrass plant used as the "stressor" in this study has been modified to increase triacylglyceride levels in foliage by 100% to increase the metabolisable energy content of forage for grazing animals. The high-ranking "test" species chosen to illustrate the concept are New Zealand native species Wiseana cervinata (Walker) (Lepidoptera: Hepialidae), Persectania aversa (Walker) (Lepidoptera: Noctuidae), and the self-introduced grey field slug, Deroceras reticulatum (Müller).