Evaluation of the environmental and agricultural potential of arid lands is often limited by the lack of information on soil surface roughness and water content. The current study proposes an efficient method to retrieve these parameters of bare soil from single-channel ERS-1, -2 synthetic aperture radar (SAR) data. New equations were derived by combining the model for vertically co-polarized mode backscattering coefficient the model for the real part of dielectric constant and the empirical equation interrelating parameters of roughness. These equations allowed for calculation of the root mean square (RMS) height h of small surface roughness (h ≤ 1 cm) for naturally sandy, flat areas of the Negev desert (Israel) during dry periods when is extremely low and generally known. As soil roughness was found to be sufficiently constant under the arid environment, this study showed that calculated h could be reliably used to retrieve during the wet period. Statistical analysis of the relative errors of retrieved h and showed their high independence on the absolute values. Retrieved values of h and obtained from ERS-2 SAR data showed acceptable correlation with the direct ground measurements. Therefore, the effectiveness of the proposed methodology for h and retrieval was proved.