Image fusion or the merging of images recorded at different spatial resolutions is widely applied in various fields of remote sensing. Often the aim is to increase the visual content of an image as the result of merging low-spatial resolution-high spectral resolution imagery with a high spatial resolution image. The effect that may be expected from such cosmetic image improvements was quantified by calculating image statistics and variograms of original, 30 m spatial resolution Landsat Thematic Mapper (TM) images and spatially degraded image products with 90m and 150 m spatial resolution. The analysis shows that increasing the spatial resolution does not affect the mean and the median of the image histograms, however the standard deviation decreases and the spread of the image histogram decreases. The rate of change of the spread is dependent on the spatial structure of the image which is reflected in the variogram. For homogeneous data (e.g. data with little spatial continuity reflected in a short value of the variogram range parameter), the spread decreases more rapidly than for heterogeneous data (e.g. data with high spatial continuity reflected in long range values). Inversely it can be concluded that image fusion aiming at improving visual content and interpretability will be more successful in the case of homogenous data than for heterogenous data.