The analysis of data from surface survey continues to be dominated by approaches which make a distinction between 'sites', typically defined as high frequencies or clusters of artefacts, and the background scatter, everything else. These approaches include those based on methods of classification which solely consider the frequency of artefacts recovered from a particular location and those based on spatial statistics which consider the clustering or dispersion of the locations from which artefacts were recovered. All too often, the resulting analysis is uncritical and fails to take into account the complexity of the spatial patterning in the distribution of artefacts from surface contexts. Data from surface survey is typically aggregated by areal unit, whether by grid square, transect/spit or plot of land, and contrary to popular belief it is the areal unit rather than the artefact or site which is the basic unit of analysis. A network based approach to the analysis of the spatial distribution of artefacts from surface contexts is proposed in this paper which treats the areal unit as the basic unit of analysis and considers the relationships between areal units which yielded artefacts. The analytical potential of this approach is highlighted with reference to a case study from northern England and the resulting analysis is compared against that from traditional approaches based on methods of classification or spatial statistics.