The successful application of passive microwave sensors requires signatures for the unambiguous inversion of the remote sensing data. Due to the large number of object types and large variability of physical properties, the inversion of data from land surfaces is a delicate and often ambiguous task. The present paper is a contribution to the assessment of multi-frequency passive microwave signatures of typical objects on land in winter. We discuss the behaviour of measured emissivities at vertical and horizontal polarization over the frequency range of 5 to 100 GHz (incidence angle of 50 degrees) of water and bare soil surfaces, grass and snowcovers under various conditions. These data and their variabilities lead us toward a classificaion algorithm for some, but not all object classes. Most snowcovers can easily be discriminated from other surfaces, difficulties occur for fresh powder snow if 94 GHz data are not available. The problem of wet snow has found a solution by using a certain combination of observables.