BACKGROUND AND AIM: This symposium will focus attention on what characteristics might make studies have more impact on policy. It depends on the context – in my experience if the goal is to set an exposure limit, either a no adverse effect level or regulatory limit, then a few “key studies” tend to emerge. These are studies which stand out as having credible quantitative exposure assessment, good power, and show effects or thresholds at the lowest levels, along with other good design qualities. For hazard assessment more of the literature contributes and this talk aims to consider the place of diverse literature designs, feeding into standard setting for PFAS (perfluorinated compounds). METHODS: Research on PFAS has exploded in recent years and because of their toxicity and biopersistence many regulators have sought to set exposure limits for the most significant. This talk reviews the design options for identifying risks using the example of PFAS, illustrated with examples from recent studies. RESULTS:There is wide variation in the choice of key effect or value of exposure limit between different regulators. Different commentators judge the health impacts of PFAS very differently, and this may be understood in terms of different sensitivities to different biases. For example, biomarker studies are persuasive because they have stable individual exposures or unconvincing because biomarker levels reflect personal determinants which may confound. Modelled exposures are persuasive because they avoid individual confounding but models introduce model uncertainty. Geographic studies avoid the above two problems but may be vulnerable to ecologic bias. Where the association is present across these multiple designs, as in the case of cholesterol associations with PFOA, the presence of hazard is persuasive. CONCLUSIONS:All observational studies have methodological limitations, but in the case of PFAS at least, integration of evidence across diverse designs, or “triangulation” is a powerful tool in hazard assessment. KEYWORDS: Environmental epidemiology, Policy, Causal inference, PFAS