<jats:sec><jats:title>Background:</jats:title><jats:p>To improve the use of evidence in policy and practice, many organisations and individuals seek to promote research-policy engagement activities, but little is known about what works.</jats:p></jats:sec> <jats:sec><jats:title>Aims and objectives:</jats:title><jats:p>We sought (a) to identify existing research-policy engagement activities, and (b) evidence on impacts of these activities on research and decision making.</jats:p></jats:sec> <jats:sec><jats:title>Methods:</jats:title><jats:p>We conducted systematic desk-based searches for organisations active in this area (such as funders, practice organisations, and universities) and reviewed websites, strategy documents, published evaluations and relevant research. We used a stakeholder roundtable, and follow-up survey and interviews, with a subset of the sample to check the quality and robustness of our approach.</jats:p></jats:sec> <jats:sec><jats:title>Findings:</jats:title><jats:p>We identified 1923 initiatives in 513 organisations world-wide. However, we found only 57 organisations had publicly-available evaluations, and only 6% (141/2321) of initiatives were evaluated. Most activities aim to improve research dissemination or create relationships. Existing evaluations offer an often rich and nuanced picture of evidence use in particular settings (such as local government), sectors (such as policing), or by particular providers (such as learned societies), but are extremely scarce.</jats:p></jats:sec> <jats:sec><jats:title>Discussion and conclusions:</jats:title><jats:p>Funders, research- and decision-making organisations have contributed to a huge expansion in research-policy engagement initiatives. Unfortunately, these initiatives tend not to draw on existing evidence and theory, and are mostly unevaluated. The rudderless mass of activity therefore fails to provide useful lessons for those wishing to improve evidence use, leading to wasted time and resources. Future initiatives should draw on existing evidence about what works, seek to contribute to this evidence base, and respond to a more realistic picture of the decision-making context.</jats:p></jats:sec>