We conduct a CV and a CE experiment using a generic rather than a situation-specific study design in order to obtain a generic marginal value function for different types of natural areas with different characteristics in the Netherlands. We develop a modelling approach in which we use CV and CE choice data in one model. The value function obtained shows that people attach value to the presence of natural areas, and that these values vary due to differences in the magnitude of areas, in distance to areas, and differences in accessibility and fragmentation of natural areas. We discuss how the value function can be used to incorporate potential substitution between natural areas. We also show that for approximately one-fifth of our sample there is no added value having any kind of natural area nearby and that people living in small municipalities tend to be indifferent between accessible and inaccessible areas. Our approach produces a generic marginal value function, which may be used to inform cost–benefit analyses and land use planning decisions that wish to incorporate positive and negative welfare externalities related to changes in the provision of natural areas. However, we also show that applying this value function to the regional or local scale may produce biased value estimates.