1 and the pharmaceutical industry 2 provide their perspectives on the new 2017 EMA guideline on nonclinical and clinical aspects of first-in-human (FIH) and early clinical trials. This guideline was a response of the EMA to the tragic outcomes of the FIH trial with the fatty acid amide hydrolase (FAAH) inhibitor, BIAL 10-2474, in 2016. The guideline puts more emphasis on the better use of preclinical data to guide rational dose selection of FIH studies. The EMA and the pharmaceutical industry appear to be in agreement on how this applies to selecting a safe starting dose, but their perspectives seem to diverge when it comes to defining the pharmacodynamic range and maximum dose that can be explored in an FIH study. 1,2 The industry perspective 2 states that "…the predicted therapeutic range at the time of FIH conduct is based on translational PK/PD (pharmacokinetic/ pharmacodynamic) data, which in many cases is adjusted on emerging clinical data" and "…the exposure maximum is typically defined by mean animal exposure." This view, as well as the new EMA guideline, seems to suggest that FIH doses can only be estimated on the basis of preclinical data and appear to ignore the potential role of mechanistic model-based approaches, such as quantitative systems pharmacology (QSP), which may or may not use preclinical data. This point is particularly relevant for the BIAL 10-2474 case, in which an available QSP model could have been used to set stopping criteria well below the (fatal) highest dose tested based on understanding of the pharmacodynamic range. Two years before the BIAL 10-2474 trial, we published a QSP model 3 that explained why the selective FAAH inhibitor, PF-04457845, had demonstrated a limited pharmacodynamic range in terms of modulation of the key biomarker, anandamide, in an FIH trial. On the basis of this, the model predicted a limited modulation of the target of interest (Cannabinoid CB1 receptors) in the brain, the magnitude of which would saturate at relatively low doses of PF-04457845; and provided a rationale for the failed phase II proof-of-concept study in patients with osteoarthritis 3 and confidence that there was no rationale to test a higher dose. Critically, in the absence of relevant preclinical animal models of pain, the QSP model was entirely based on and calibrated against in vitro and (human) literature data.On the basis of similarity of the biomarker data presented by the BIAL 10-2474 team 4 compared with the PF-04457845 FIH, we propose that the higher doses tested in their study exceeded the required pharmacodynamics range to test the FAAH hypothesis (as defined by the QSP model) and that the severe adverse events were attributable to expression of additional pharmacology through other mechanisms. 5 In our view, use of an existing QSP model could have provided a more meaningful prediction of the pharmacodynamic range and maximum dose for the BIAL 10-2474 FIH than preclinical animal data and arguably might have prevented the disastrous outcome of the trial. The use of such mechanisti...