Many of the resources developed to promote the use of evidence in policy aspire to an ideal of rational decision making, yet their basis in the decision sciences is often unclear. Tracing the historical development of evidence-informed policy to its roots in "evidence-based medicine" (EBM), we distinguish between two understandings of how research evidence may be applied. Advocates for EBM all seek to use research evidence to optimise clinical care. However, some proponents argue that 'uptake' of research evidence should be direct and universal, for example through wide-scale implementation of 'evidence-based practices'. In contrast, other conceptualisations of EBM are rooted in "expected utility theory," which defines rational decisions as choices that are expected to result in the greatest benefit. Applying this theory to medical care, clinical decision-making models clearly demonstrate that rational decisions require not only a range of relevant evidence, but also expertise to inform judgments regarding the credibility of estimates and to assess fit-to-context, and stakeholder preferences and values to weigh trade-offs among competing outcomes. Using these models as exemplars, we argue that attempts to apply research evidence directly to practice or policy without consideration of expert judgement or preferences and values reflect fundamental misconceptions about the theory of rational decision making that can impede implementation. In turn, the decision sciences highlight the need to consider the role of expertise and judgment when interpreting research evidence, the role of preferences and values when applying it to specific decisions, and the practical limits imposed by the uncertainty inherent in each.