To test this hypothesis we used two contrasting methodologies. First, a series of satellite images—Landsat TM and ETM+ from August 1984, 1991 and 2001—was processed to obtain a time series of the Normalized Vegetation Index (NDVI) for the whole study area of about 2000 km2. Since the NDVI is a continuous variate, changes in the heterogeneity of the vegetation activity were assessed by a spatial variance filter. Second, a time series of more detailed aerial photographs from a smaller area were classified into vegetation types, and the time variation of several landscape indexes was analyzed. Although the nature of the variables and the scale and spatial resolution of the analyses differ, the results from both methods supported the hypothesis about the effect of the vegetation succession process on the landscape heterogeneity. Understanding the driving factors of natural vegetation recovery in complex landscapes can be of great interest to foresee the effects of land use change in many similar mountain areas of the World, specially in the developed countries in which the abandonment of marginal farm land has become a common process.