Land cover diversity is often used as a surrogate of habitat heterogeneity. Nevertheless, its spatial pattern has received limited attention. Here, we examine Corine land cover diversity patterns across Europe, and test (1) if geographical (longitudinal, latitudinal) gradients exist, (2) if the scale of analysis (and specifically the grain of analysis) influences the patterns, and (3) if the thematic resolution affects the results. We estimated diversity landscape metrics for 2818 locations throughout Europe. We analysed the spatial pattern at five grains (0.25, 1, 25, 100 and 625 km2), and for three hierarchical levels of the Corine Land Cover 2000 classification scheme. To account for spatial autocorrelation, we used Clifford’s test. Latitude was significantly correlated with land cover diversity (once spatial autocorrelation was taken into account) only at large grain sizes. Longitude, with a few exceptions at fine grain, was not correlated to land cover diversity. The spatial pattern of land cover diversity is scale-dependent, with spatial pattern at fine grain (<1 km2) being statistically independent of the pattern at large grain (625 km2). Also the grain of the analysis affected the spatial autocorrelation of land cover diversity. Fine grain analysis displayed autocorrelation, but over short (hundreds of kilometres) distances, while large grain analysis displayed autocorrelation over longer distances (thousands of kilometres). Increasing the detail of the thematic resolution seems to have effects similar to increasing the grain size. The thematic resolution in certain cases influenced the results qualitatively and thus inference from low-resolution landscape analysis should be done with caution.