SPOT XS oblique data are compared with Landsat Thematic Mapper (TM) vertical data at two different times. Under the assumption that analogous spectral bands are equivalent, the differences in spectral responses observed over land cover classes are attributed to the effects of angular anisotropic reflection. This article shows that land cover classes cannot easily be characterized by their anisotropy. The magnitude of anisotropy varies strongly with the radiometric processing levels (i.e., absolute calibration or normalization) but has no significant impact on the overall classification results. The normalization method based on the use of pseudoinvariant objects is sufficiently effective as to correct most anisotropic disturbances and should facilitate an alternative use of SPOT and TM data.