The monitoring of desertification processes, and particularly the development of "early-warning" systems, is an increasingly important development in the management of drylands. It has been shown that the patch size distribution of dryland vegetation can be described using power laws and that deviations from such patterns may be used as an early-warning signal for the onset of desertification. We tested this idea using data from 29 semiarid steppes located along a latitudinal gradient in Spain. A truncated power law (TPL) fitted the patch size distribution of perennial vegetation better than a power law in all the evaluated sites. Variations in this distribution, as measured with the scaling exponent (gamma) of the TPL, were not related to total perennial cover, but a negative logarithmic relationship was found between gamma and soil variables related to desertification processes (total nitrogen, total phosphorus, and organic carbon). A positive and stronger linear relationship was found between total perennial cover and the same soil variables. Our results suggest that deviations from a patch size distribution characterized by a power law are not directly related to desertification. They also indicate that plant cover can be used to effectively monitor key variables linked to desertification processes.