In this study, vegetation optical depth from ESA's Soil Moisture and Ocean Salinity (SMOS) satellite mission is combined with other existing remote sensing, meteorological and literature data in order to obtain values of gravimetric vegetation water content (M_g). The methodology combines an effective medium model valid at passive microwave frequencies with a vegetation dielectric constant model. The algorithm is calibrated for 11 global vegetation classes. The resulting product consists of temporally dynamic ~25 km global grids of M_g. The first maps clearly show seasonal differences in vegetation water, which vary for the different continental regions due to variations in e.g. latitude, climate and landcover type. This new vegetation water product is unique and offers important complementary information to existing vegetation indices.