Quantitative reconstruction of the area cleared of forest in the past is essential to assess the possible indirect anthropogenic impacts on the past environment of Europe, including past climate. We apply a simul ation model of pollen dispersal and deposition (1) to re-examine the relationship between pollen and landscape openness, often uncritically inferred from non-arboreal pollen (NAP) percentages alone, and (2) to predict the relevant source area of pollen, the smallest spatial scale of vegetation that can be reconstructed from pollen records. The simulations use landscapes simplified from the modern open agricultural and semi-open forested regions in southern Sweden where traditional cultural landscapes still remain. The model is appropriate, because the simulated pollen assemblages resemble the pollen assemblages observed in each of the two landscape types, and because the simulated relationships between NAP percentages and percentage cover of open land within 1000 m agree with the empirical relationships. The simulated relevant source area of pollen is the area within 800–1000 m from both small hollows and 3-ha ponds. NAP percentages give only a rough first approximation of the percentage cover of open land. More comprehensive methods will be required to obtain quantitative estimates of open land from fossil pollen.