The weight-of-evidence, case-by-case approach is considered the most appropriate way of assessing the allergenicity of genetically modified (GM) food and feed. This scientific opinion discusses various aspects to increase the strength and accuracy of this approach, including the latest developments pertaining to clinical aspects of allergic reactions, structural aspects of GM food and feed and in silico approaches, as well as IgE binding studies and cell-based methods, profiling techniques and animal models. In this context, conclusions and recommendations are provided to update and complement current risk assessment strategies for the allergenicity assessment of newly expressed protein(s) and whole GM food and feed. In summary, it is recommended that with regard to the search for sequence homology and structural similarities, the local alignment method with a known allergen with a threshold of 35% sequence identity over a window of at least 80 amino acids is considered a minimal requirement. When IgE binding tests are considered necessary, e.g. when there is sequence homology and/or structure similarity with known allergens, the use of individual sera from allergic individuals rather than pooled sera is recommended. In addition to the pepsin resistance test, it is recommended that the resistance to digestion of the newly expressed proteins is evaluated using other in vitro digestibility tests mimicking physiological conditions of humans. Finally, when the recipient of the introduced gene is allergenic, in order to compare the allergenicty of the whole GM plant with that of its appropriate comparator(s), it is recommended that relevant characterised endogenous allergens are included in the comparative compositional analysis of the GM plant and its appropriate comparator(s). Proposals for the use of additional testing that may improve the weight-of-evidence approach and suggestions for further evaluation of new promising methods that are as yet in an early phase of development are also addressed.