BACKGROUND: Discussions within the health community routinely emphasise the importance of evidence in informing policy formulation and implementation. Much of the support for the evidence-based policy movement draws from concern that policy decisions are often based on inadequate engagement with high-quality evidence. In many such discussions, evidence is treated as differing only in quality, and assumed to improve decisions if it can only be used more. In contrast, political science scholars have described this as an overly simplistic view of the policy-making process, noting that research 'use' can mean a variety of things and relies on nuanced aspects of political systems. An approach more in recognition of how policy-making systems operate in practice can be to consider how institutions and ideas influence which pieces of evidence appear to be relevant for, and are used within, different policy processes. METHODS: Drawing on in-depth interviews undertaken in 2015-2016 with key health sector stakeholders in Cambodia, we investigate the evidence perceived to be relevant to policy decisions for three contrasting health policy examples, namely tobacco control, HIV/AIDS and performance-based salary incentives. These cases allow us to examine the ways that policy-relevant evidence may differ given the framing of the issue and the broader institutional context in which evidence is considered. RESULTS: The three health issues show few similarities in how pieces of evidence were used in various aspects of policy-making, despite all being discussed within a broad policy environment in which evidence-based policy-making is rhetorically championed. Instead, we find that evidence use can be better understood by mapping how these health policy issues differ in terms of the issue characteristics, and also in terms of the stakeholders structurally established as having a dominant influence for each issue. Both of these have important implications for evidence use. Contrasting concerns of key stakeholders meant that evidence related to differing issues could be understood in terms of how it was relevant to policy. The stakeholders involved, however, could further be seen to possess differing logics about how to go about achieving their various outcomes - logics that could further help explain the differences seen in evidence utilisation. CONCLUSION: A comparative approach reiterates that evidence is not a uniform concept for which more is obviously better, but rather illustrates how different constructions and pieces of evidence become relevant in relation to the features of specific health policy decisions. An institutional approach that considers the structural position of stakeholders with differing core goals or objectives, as well as their logics related to evidence utilisation, can further help to understand some of the complexities of evidence use in health policy-making.