Thirteen surface moss samples were collected for pollen analysis from an area of heathland in western Norway. Vegetation composition at different distances around the sampling locations was measured using three different survey methods; rooted frequency within a sub-divided 1 m x 1 m quadrat, visual estimates of cover within a 1 m x 1 m quadrat and a modified form of the 'circle-walking method'. Extended R-value analysis was used to explore the pollen-vegetation relationships for five main taxa, Calluna vulgaris, Vaccinium-type, Cyperaceae, Poaceae and Potentilla-type. The estimates of relevant source area of pollen obtained were similar regardless of the vegetation survey method. Values obtained were always under 4 m. However, estimates of relative pollen productivity and the background pollen component (proportion of pollen coming from vegetation growing beyond the relevant source area of pollen) differ markedly depending on the method of vegetation survey chosen. This has important implications for the quantitative reconstruction of past vegetation cover.