The Antarctic Ice Sheet (AIS) is likely to be a major contributor to future sea-level rise, which drives the need for a thorough understanding of ice-sheet mass changes. However, estimates of ice mass changes over recent decades from satellite gravimetry (GRACE and GRACE-FO or just simply GRACE) and satellite altimetry (e.g. CryoSat-2) differ strongly in several regions of Antarctica and have large uncertainties. GRACE has the great advantage of being ultimately sensitive to mass changes on a monthly scale, but is limited by spatial resolution (a few hundred kilometres), correlated noise and—especially in Antarctica—uncertainties introduced by the application of the Glacial Isostatic Adjustment (GIA) correction. CryoSat-2 captures ice mass changes at high spatial and temporal resolution, but suffers from assumptions about volume-to-mass conversion and correlated errors in space and time resulting from the altimetry-data processing chain, e.g. time-variable radar-signal penetration into snow and firn. Here we present results from a method that is able to exploit the advantages of GRACE and CryoSat-2 while overcoming their limitations. The inverse approach allows co-estimation of GIA instead of using GIA model output, and incorporates results from regional climate modelling and firn modelling to improve the volume-to-mass conversion for satellite altimetry. The mass balance time series are compared with data from ICESat-2 and estimates from the Ice sheet Mass Balance Inter-comparison Exercise (IMBIE). We also discuss perspectives of satellite data combination with future missions such as MAGIC and CRISTAL.