Environmental chemicals, including endocrine disrupting chemicals (EDCs), pose a threat to human health. Actions are taken by scientists, assessors, regulators, and policymakers around the world to improve testing strategies for chemical substances, including pushing towards greater reliance on data from new approach methodologies to replace animal toxicity studies. This paradigm shift is envisioned to ultimately replace animal testing altogether for many purposes. As regards identification and regulation of EDCs, this poses certain challenges in that current guidelines—at least within the European regulatory framework—stipulate that adverse outcomes are to be demonstrated in an intact organism. The new testing paradigm is, of course, to find ways of dealing with this dilemma. However, another challenge still remains, even if the “intact organisms” definition changes or is replaced, namely the challenge of predicting apical adverse effects resulting from endocrine disruption. The adverse outcome pathway (AOP) framework provides a good platform for identifying and regulating EDCs based on both non-animal and animal (or human) data, but also here we are confronted with the same challenge: how to predict adverse effects in complex organism from simple test assays that are based on reductionist principles? In this article, the challenge of “emergent properties” in predictive toxicology is highlighted as a cautionary footnote because, although a future relying far less on animal toxicity testing is both desirable and sensible, the pace at which we transition to the new paradigm should ensure that human health, and the environment, is safeguarded from harmful chemical substances.