The SMOS project (CNES/ESA) aims at developing a L-band interferometric radiometer able to provide global estimates of surface soil moisture (ws) with a sampling time of 2-3 d. Several studies showed that the assimilation of ws time series enable one to retrieve the root-zone water content, provided the atmospheric forcing is available from ancillary information. Two real cases of variational assimilation of ws are presented. A simple iterative optimization of the direct model is applied to the ISBA (Interactions between Soil, Biosphere, and Atmosphere) surface scheme. The next step is the direct assimilation of brightness temperatures as likely to be observed from space. A first example of soil moisture and vegetation characteristics analysis using L-band radiometric measurements and the ISBA model is presented. A field campaign was set in Southwestern France in order to validate such an approach using a long enough data-set.