Hazard identification is the first step in assessing the risk of a genetically modified (GM) crop. It employs the concept of substantial equivalence to evaluate crop safety. The current process relies on subjective opinions to integrate various comparisons among the GM crop, the non-GM counterpart and an assortment of non-GM references over an array of key endpoints measured in field trials. The pre-eminent need to control the consumer's risk in hazard identification has been left unaddressed. The current paper develops statistical strategies to resolve this issue. Hypotheses of individual tests are explicitly defined to reflect the study objectives. They are then grouped into families and connected by logical operators according to decision rules commonly used in crop safety evaluation. This pre-specification of hypotheses arranged in an organized layout leads to a simple, transparent decision-making process where the consumer's risk can be managed directly. A two-stage multiplicity adjustment procedure is created by applying fundamental principles for multiple testing to the newly assembled families of hypotheses. The practical utility of the proposed procedure is shown in a real-world example. Besides being easy to implement and convey, the proposed statistical strategies accommodate the addition of supportive evidence for safety and allow the nature of the genetic modification to be taken into account.