In order to assess the retrieval of soil moisture from ERS 1 (European Remote Sensing Satellite) synthetic aperture radar (SAR) data, an inversion procedure based on the integral equation model (IEM) <Fung et al., 1992> is developed. First, the IEM is used to analyze the sensitivity of radar echoes (in terms of the backscattering coefficient &sgr;0) to the surface parameters (roughness and dielectric constant) under ERS 1 SAR configuration. Results obtained for random rough bare soil fields show that the effect of surface roughness is very strong, particularly in the case of smooth surfaces, and that the sensitivity of &sgr;0 to dielectric constant is independent of the radar configuration and the roughness conditions. This means that the range of variation of backscattering with respect to the dielectric constant variation of dry to wet soil remains the same (about 5 dB) for any roughness condition and radar configuration. The possibility of applying the inversion procedure to retrieve soil moisture is investigated using a set of data collected in a test site situated near Naples, Italy, during the Sele Synthetic Aperture Radar experiment (SESAR) campaign (November 1993). Simultaneous with ERS 1 overpasses, dielectric constant and roughness measurements were taken over two flat bare fields. From this analysis it is found that the inversion of backscattering from ERS 1 SAR into soil moisture is not reliable without accurate information on roughness if the surface is smooth. In this case it is observed that the sensitivity to the roughness parameters is much higher than the sensitivity to dielectric constant, so that even a small error in the measurement of this parameter can affect the retrieved value of soil moisture significantly. The inversion procedure provides more reliable soil moisture estimates when surfaces rougher than those analyzed in the field experiment are considered. ¿ American Geophysical Union 1996