Vegetation crop marks may be formed in areas where vegetation overlays near-surface archaeological remains. These features retain soil moisture with different percentage of moisture compared to the rest of the crops of an area. Depending on the type of feature, crop vigour may be enhanced or reduced by buried archaeological features. Satellite imagery has been already applied successfully in several archaeological investigations for the detection of buried archaeological features based on such crop marks. However, such features can only be classified when their spectral characteristics are different from their surroundings. Difficulties might occur when spatial resolution (pixel size) of the satellite sensor is low enough in order to distinguish crop marks from their surroundings. In these cases up-scaling techniques, like linear spectral un-mixing (LSU), can be used in order to improve spatial resolution and to enhance image results. The aim of this paper is to assess LSU technique for the detection of archaeological sites. LSU is based on the assumption that within a given scene, the surface is dominated by a small number of distinct materials that have relatively constant spectral properties (called endmember). LSU technique was evaluated at several Neolithic tells (magoules) located at the Thessalian plain. Different multispectral satellite images (mainly Landsat TM/ETM+) have been used for this purpose. The final results were compared with other standard remote sensing techniques like Principal Component Analysis, vegetation indices, Tasseled Cap and ground spectroradiometric data.