In the risk assessment process, hazard characterization pivots on the mode(s) of action of a toxicant and on the possibility to define the relationships between the doses and the biological responses, with the aim of setting a ‘safe dose’. This may be defined as a pragmatic threshold to be used in risk management, whereas the definition of a ‘real threshold’ is burdened by uncertainties. As regards endocrine disrupting chemicals (EDCs), a long and intensive discussion has taken place on a number of topics. Non-monotonic dose-responses do exist and probably reflect the existence of qualitatively different mechanisms, each giving rise to a dose–response curve for the relevant effects. Indeed, EDCs may regulate hormone pathways in different ways at different concentrations at a target or elicit other toxicological mechanisms at dose levels higher than those causing endocrine effects. The low-dose issue is a poorly defined one. Investigating EDCs using ad hoc endpoints not considered in regulatory studies and/or during vulnerable windows may plausibly detect adverse effects at dose levels where no other toxic outcomes were observed. Also, high-concern chemicals may cause adverse effects at actual environmental exposure levels, before proper regulatory and risk management measures are implemented. Thus, experimental studies investigating sensitive endpoints can actually observe adverse effects at environmentally relevant dose levels. Finally, the chapter discusses the use of adverse outcome pathways (AOP) to identify the relevant upstream events for which dose–response curves have to be sought within the framework of a mechanism- and biology-driven hazard characterization. Indeed, the use of AOP may reduce the risk of ‘drowning in complexity’.