Air quality in China is a major public health, social and economic concern. Air pollution governance and research in China have been increasingly active in the past decade, especially since 2013 when strict emission controls were implemented. Such emission control policies have been informed through dialogue between scientists and policy-makers on the sources and transport of air pollution in order to identify potential control measures. However, the process of making regulatory decisions about air pollution controls at this science-policy interface in China has rarely been analysed or discussed. We outline four classical science-policy models for making regulatory decisions proposed by scholars: (i) the decisionist model – whereby policy dictates what science and regulatory decisions are required; (ii) the technocratic model – where science dictates policy directly; (iii) the inverted decisionist model (where scientists advise policy-makers on what policy is required); and (iv) the co-evolutionary model (where policy-makers and scientists jointly create regulatory decisions). Boundary-actors play a key role in this co-evolutionary model. They operate as ‘gate-keepers’ between scientists and policy-makers. Most contemporary studies of the science-policy interface argue that the co-evolutionary model best captures the reality of how science and policy interact effectively to make regulatory decisions. To assess which of these models most closely resemble decision-making at the air pollution science-policy interface, we conducted a case study on “air quality climate services” and held workshops with Chinese scientists, decision-makers and stakeholders. A typology of existing scientific approaches to explore air quality climate science is presented. The workshop results show that the current air quality climate science-policy interplay occurs most strongly in accordance with the co-evolutionary model whereby the Beijing Climate Centre and the National Environmental Monitoring Centre operate as the key boundary actors between science and policy, specifically for a seasonal air pollution haze outlook service. We illustrate that current seasonal haze outlooks carefully avoiding quantification. We then present a conceptual framework of the air pollution science-policy interface in China, which captures the main participants and the interactive flow of information between them.