Abstract: As part of the regulatory approval process for commercial cultivation of geneticallymodified crops in the EU applicants have to asses the potential adverse effects that GM cropsmay have on human and animal health and the environment. This includes an assessment ofpotential adverse effects on non-target organisms arising from intended and unintended results ofthe genetic modification. The methodology to be used for the environmental risk assessment ofGM crops to non-target organisms has been a major subject of debate for many years. However,after much research, a conceptual framework based on a tiered approach is now widely acceptedby risk assessors, regulators and the scientific community. This methodology works well whenthe assessment is aimed at establishing the risk associated with intended effects of the geneticmodification or unintended effects that have been identified during the comparative assessment.There the transformed plant is grown alongside its conventional counterpart and a number ofplant characteristics are measured and compared. However, the regulatory process in the EU nowalso considers the requirement to assess the risk of potential unintended effects that have not beenidentified during the comparative assessment. This represents a major challenge for risk assessorsin that there is no clear basis in which to set testable hypothesis. This paper discusses some of theproblems encountered by risk assessors when trying to fulfill this regulatory requirement andexplores ways forward.