The regulatory framework for pesticides and for classification of chemicals as carcinogens, foresees that a substance that is proven to be carcinogenic in animals, may still be placed on the market provided a demonstration is made that this carcinogenicity does not verify in humans. This provision is based on two regulatory-scientific notions that were placed at the heart of European laws regarding chemicals and pesticides at the end of the 2000s, and still apply : that of « mode of action » and of « human relevance ». In short, if the biochemical mode of action of a susbtance is specific to animals and not verified in humans, it may be accepted. The notions of MOA and of human relevance have emerged as evaluative concepts in the US, in the context of controversies surrounding the EPA’s reliance on dose characterization and applicatoin of safety or uncertainty factors. These concepts were further articulated in an international policy venue, namely in the International Program for Chemical Safety of the World Health Organization, under the impetus of EPA staffers and representatives of the agrochemical industry, both American and European. Once normalized, the methodology has been adopted by the EU, most likely to provide flexibility during the negotiation of the new Regulation 1107/2009, then emerging as a more stringent « hazard-based » regime that the industry broadly opposed. This paper tests the notion of global epistemic capture, to make sense of whether and how the international political economy of pesticides reflects in the actual European regime of pesticide assessment, in both its epistemic and legal dimensions.