In this paper we evaluate data obtained during the VALEM (Validation of Land Surface Emissivity Models) experiment, which took place in September 1998 near Ludwigslust in north-eastern Germany. The emphasis of the e eld campaign was to generate datasets to validate land surface emissivity models in the microwave spectral range. We describe the approach to derive land surface emissivities from the observed data and compare observed and simulated land surface emissivities in order to further validate the Free University of Berlin’s newly developed microwave emissivity model. This model explicitly accounts for any given type of structured surfaces by splitting the surface into large-scale facets and a small-scale surface roughness. A model description can be found in previous work (PAAPE et al., 2000) so that in the current paper we only present a brief overview on the model structure and concentrate on its validation at frequency ranges between 24 and 157 GHz. The range of variability of the observed surface emissivity is about 18% (0.80‐0.98) for the different frequencies and test sites. Comparing simulated and observed nadir emissivities we e nd deviations between 0.9% and 2.5% dependent on frequency. The deviations are in the range of uncertainty introduced by the various input parameters into the emissivity model. Comparing model results with radiometer observations at different incidence angles, we e nd that the model realisticallyresembles the variation of the surface emissivity with zenith angle at different frequencies. Zusammenfassung In diesem Artikel zeigen wir Ergebnisse der Validierung des Modells der Freien Universit¨