Background: Many different concepts have been invoked in the literature to describe coordination between scientific and policy practices. However, a framework to understand different types of this boundary management is lacking, which allows 'linear' models and 'gap' metaphors to persist. Objectives: To operationalise the multidimensional 'proximity' approach from innovation studies in the context of coordination work between science and policy and employ it in a case study of hybrid research spaces in the port city of Rotterdam. Method: Semi-structured interviews and focus groups with scientists, policymakers and policy researchers involved in collaborative knowledge labs. Findings: Different proximity dimensions can both stimulate and impede productive interactions in collaborations between science and policy actors. There is no uniform mechanism to overcome coordination problems and different knowledge labs use different combinations of proximities. Discussion and conclusions: We conclude that the concepts of cognitive, social, organisational and geographic proximity enable a fine-grained articulation of the balancing act required for successful and enduring collaborative knowledge work between scientists and policymakers. The operationalisation of proximity for structured science-policy interactions led us to the insight that an analytical distinction can be made between "ordering" and "operating" enactments of proximity. Ultimately, we interpret this as two types of politics of knowledge production, and call on policymakers, researchers, and other involved parties to take this into consideration when organising structural knowledge collaborations between research and policy practices.