Arthropods form a major part of the biodiversity in agricultural landscapes. Many species are valued because they provide ecosystem services, including biological control, pollination and decomposition, or because they are of conservation interest. Some arthropods reduce crop yield and quality, and conventional chemical pesticides, biological control agents and genetically engineered (GE) crops are used to control them. A common concern addressed in the ecological risk assessment (ERA) that precedes regulatory approval of these pest control methods is their potential to adversely affect valued non-target arthropods (NTAs). A key concept of ERA is early-tier testing using worst-case exposure conditions in the laboratory and surrogate test species that are most likely to reveal an adverse effect. If no adverse effects are observed in those species at high exposures, confidence of negligible ecological risk from the use of the pest control method is increased. From experience with chemical pesticides and biological control agents, an approach is proposed for selecting test species for early-tier ERA of GE arthropod-resistant crops. Surrogate species should be selected that most closely meet three criteria: (i) Potential sensitivity: species should be the most likely to be sensitive to the arthropod-active compound based on the known spectrum of activity of the active ingredient, its mode of action, and the phylogenetic relatedness of the test and target species; (ii) Relevance: species should be representative of valued taxa or functional groups that are most likely to be exposed to the arthropod-active compound in the field; and (iii) Availability and reliability: suitable life-stages of the test species must be obtainable in sufficient quantity and quality, and validated test protocols must be available that allow consistent detection of adverse effects on ecologically relevant parameters. Our proposed approach ensures that the most suitable species are selected for testing and that the resulting data provide the most rigorous test of the risk hypothesis of no adverse effect in order to increase the quality and efficiency of ERAs for cultivation of GE crops.