Aboveground biomass (AGB) is an essential component of the Earth's carbon cycle. Yet, large uncertainties remain in its spatial distribution and temporal evolution. Satellite remote sensing can help improve the accuracy of AGB estimates. In particular, the L-band (1.41 GHz) vegetation optical depth (VOD) derived from the SMOS (Soil Moisture and Ocean Salinity) mission is a good AGB proxy. Averaging the SMOS L-VOD over a year and linking it to an existing AGB map constitute a well-established method to derive a spatial relationship between the two quantities. Then, a temporal extrapolation of this spatial relation derives global and harmonized AGB time series from the L-VOD. This study refines this protocol by analyzing the impact of three factors on the AGB–VOD calibration. First, an analysis shows that ascending and descending VOD can be properly merged to estimate the AGB. Second, the use of a single global spatial relationship is preferred over several regional ones. Third, this new AGB dataset is compared with other published AGB datasets to assess the validity of the temporal extrapolation. The produced dataset provides vegetation biomass values up to 300 Mg ha−1 from 2011 onward. It shows more interannual variability than the other available time series and presents globally lower AGB estimates. In general, the resulting AGB is consistent with the AGB maps of the Climate Change Initiative (CCI) Biomass version 5 (average Pearson's correlation coefficient 0.87) and can be used in AGB studies. The AGB dataset has been produced from the Level 2 SMOS products with one global VOD–AGB relationship, mixing ascending and descending orbits. The AGB dataset, including the spatial bias, is open-access and the NetCDF files are available at https://doi.org/10.12770/95f76ff0-5d89-430d-80db-95fbdd77f543 (Boitard et al., 2024).