In translational medicine, research decisions are often based on complex probability statements that are subject to uncertainty. Provided that risks and burdens for humans or animals may occur or sensitive data are collected, a prospective justification of research projects is ethically necessary. At the moment, however, there is no established theoretical framework nor are there clear criteria how the quality of evidence and the predictive value of existing, often heterogeneous translational information can be captured in order to create good prospective justifications. The systematic aggregation of evidence may comprise deductive conclusions based on sets of individual preclinical studies, mechanistic considerations at the theoretical level, analogies to relevant reference classes, but also complex statistical-inductive conclusions about a larger "portfolio", "cluster" or "network" of evidence using machine learning and artificial intelligence. To some extent, it is still necessary to clarify conceptually what characterizes the quality of such epistemic and normative justifications in translational biomedical research in the first place. The Good Justification Practice project investigates how research decisions are prospectively justified in current practice (e.g. when rationales are registered or applications are submitted to funding agencies or ethics committees). In addition, concepts, criteria and recommendations are developed for an evidence-based evaluation of the prospective justification of research decisions that best implement the ethical and normative requirements. The aim is to enable the prospective evaluation of the quality of justification in translational biomedical research in a comprehensible, transparent, and evidence-based manner.