In this paper a description is given of a physically based theoretical ocean backscatter model (called the VIERS-1 model) for intermediate incidence angles, and a comparison of its performance against the CMOD4 empirical model is made. The VIERS-1 scatterometer algorithm is based on a two-scale composite surface model which includes both specular and Bragg scattering. Its short wave model is based on the energy balance equation and accounts for viscous damping, slicks, dissipation due to whitecapping, and nonlinear three- and four-wave interactions. A number of parameters in the model have been determined by means of laboratory data and analyzed European Centre for Medium-Range Weather Forecasts (ECMWF) winds. Because of the two-scale approach the wave number up to which Bragg scattering applies should be determined. This is done by means of laboratory data at X band. In addition, laboratory data of the wave spectrum have been utilized to validate the VIERS-1 short wave spectrum. An inverse of the algorithm is developed to derive wind speed and direction from the observed (ERS-1) backscatter and by comparison with ECMWF analyzed winds' three parameters for the short wave spectrum, namely, the Phillips parameter, the directional width of the spectrum, and the wave number boundary between gravity waves and short waves have been obtained. Comparisons between VIERS-1, C band model, version 4 (CMOD4), and ECMWF analyses are made. VIERS-1 performs better in the high wind speed range, and this feature is of importance when scatterometer winds are assimilated into an atmospheric model. However, in terms of backscatter rather than wind speed, CMOD4 shows better results. It is suggested that this is caused by the too simple directional distribution of the VIERS-1 short wave spectrum. ¿ 1998 American Geophysical Union