In the institutional set-up of (economic) policy preparation in the Netherlands there is ample interaction between scientific insights and policy proposals. This Dutch polder model lays much emphasis on the social dialogue to come to an agreement on, and have public support for policy proposals. It was very much the idea of Jan Tinbergen winner of the first Nobel price in economics to have a clear separation in policy preparation between (i) trying to reach consensus on the working of the economy, as formalised in econometric models; (ii) come to a compromise on policy goals between the various minority parties of the government; and (iii) rely on independent and undisputed data collection by an autonomous Central Bureau of Statistics (CBS). The aim of this separation of responsibilities is to guarantee, as much as possible, the scientific quality of policy preparation and at the same time to gain public support for policy measures so that implementation costs are kept low. This paper discusses the working of this institutional set-up, its historical background and the mechanisms of quality control and reputation which are essential for the interaction between scientific knowledge and policy preparation to remain fruitful.