Background and Aims Policymakers frequently encounter complex issues, and the role of scientists as policy advisers on these issues is not always clearly defined. We present an overview of the interdisciplinary literature on the roles of scientific experts when advising policymakers on complex issues, and in particular on the factors that influence these roles. Methods A structured literature search was combined with literature found in reference lists of peer reviewed papers (the snowball method). In total, 267 publications were analyzed using scientometrics analyses (discipline clustering analysis and co-citation analysis) followed by a qualitative analysis and interpretation. Results The scientometrics analysis shows an amalgam of disciplines that publish on our research topic. Five clusters of authors were identified based on similarities in the references used: post-normal science, science and technology studies, science policy studies, politics of expertise and risk governance. The content of the clusters demonstrates that authors in different clusters agree that the role of experts is influenced by the type of problem (simple or complex) and by other parties (the public and stakeholders). However, opinions vary on the extent to which roles can vary and the necessity to explicate different viewpoints. Discussion and conclusions Publications on scientific experts who provide policy advice affirm that such experts should and do hold different roles, depending on the type of problem and factors such as values held by the expert and the type of knowledge. We conclude that research on expert roles has remained mostly theoretical. Existing theories about science systems can be used to study real policy advice processes. Most theories are well elaborated, but empirical proof for the described changes, roles and processes is limited.