Model and corresponding algorithm able to retrieve both soil moisture and surface roughness of a bare soil from a pair of measured hh - and vv -polarised SAR images are here presented. The estimation procedure is founded on the prediction of power densities and correlation between the copolar channels provided by the Polarimetric Two-Scale Model, recently proposed by the authors. The performance of the estimation algorithm is assessed by comparing obtained retrieval results with “in situ” measurements. To this aim, data from Little Washita SMEX'03 campaign, whose results provide a popular and freely available reference, are used.