There is a broad consensus that policies, and the policy documents informing them, should be based on high-quality scientific evidence. In practice, however, authors of policy documents face two major challenges: limited access to scientific outputs and limited capacity to assess their validity. This article explores this problem and the impact of the recent exponential growth of open access publishing. Drawing on the literature on citation determinants and guided by signalling theory, we hypothesize that policy documents cite open access publications displaying quality signals, in ways that differ from academic citations. To test the hypotheses, we investigate the determinants of policy and scholarly citations for all scientific publications on climate change and global warming indexed in Scopus from 1999 to 2024, along with all policy documents citing these publications from the Overton database. We found that open access had a much greater impact on citations from policy documents than from scientific documents. At the same time, the effect on policy citations is concentrated on research produced through international collaborations, by authors affiliated with highly prestigious institutions, and published in higher-ranked journals. Hence, while expanding the availability of scientific findings on the user side, open access reinforces existing hierarchies in scientific recognition on the producer side.