摘要:Authors Blandine Doornaert , Annick Pichard Institut national de l’environnement industriel et des risques (Ineris), Unité « Évaluation toxicologique des substances chimiques » (ETSC), Département des risques chroniques (DRC), Parc technologique ALATA, 60550 Verneuil-en-Halatte Key words: epidemiology, environmental exposure, chemicals, dose-response relationship, drugs, models, theoretical, chemicals, risk assessment, toxicological reference values, toxicology Page(s) : 191-8 Published in: 2006 Accurate toxicity reference values are needed to conduct health risk assessments. Often, however, risk assessors must make a choice among several toxicity reference values (TR values) recommended by different toxicological agencies for the same substance with the same exposure route and effect. Some French agencies (DGS and InVS), working with other organizations, have developed pragmatic criteria to help risk assessors make this choice, but as far as we know no scientifically rigorous methodology has been proposed for national harmonization. Accordingly, risk assessors tend to select the most conservative value. Our objectives here are to suggest both an in-depth methodology to evaluate the scientific quality of TR values and appropriate pragmatic criteria to help risk assessors choose accurate TR values. This paper itemizes and discusses several key points to be considered in assessing the scientific quality of TR values: consistency of scientific data, critical effect, critical dose or concentration, uncertainty factors, exposure scenarios and mathematical models. At the same time, attention should be paid to the key studies and to any peer review process used by agencies to select their TR values.