Public health professionals routinely use toxicological reference values (TRVs), which characterize the relationship between human exposure to a toxic substance and the occurrence or severity of an observed adverse effect. TRVs are primarily established by international or national bodies. They are derived from observed dose-response relationships and are specific to a given effect, route of exposure, and duration of exposure. Their derivation and definition differ depending on whether a toxicity threshold is considered or whether a no-threshold effect is assumed. Animal studies are the primary source of toxicological data due to the difficulty of obtaining robust results in humans. The selection of these TRVs is a critical step in health risk assessment because, when combined with hazard identification and population exposure estimation, it enables the characterization of a risk to human health. This step is guided by adherence to selection criteria that do not, however, refer to the methods used to construct TRVs. Current trends emphasize the need to utilize available biological knowledge on each substance to improve the accuracy of TRVs. The purpose of this document is to describe and explain the theoretical and methodological foundations of TRV derivation, in order to incorporate this dimension into the selection of a TRV during a risk assessment. A better understanding of the data, their quality, and how they are generated leads to risk assessments conducted under the best possible conditions, in accordance with public health. (R.A.)