Antarctic Sea Level Anomalies (SLA) remain poorly observed due to the presence of sea ice, which hampers conventional satellite altimetry. Recent advances in lead-based retracking techniques have enabled SLA estimation within ice-covered regions. However, existing products are temporally limited, with currently available dataset covering periods shorter than 10 years. In this study, we extend the time span of SLA reconstruction by processing multiple satellite missions: Envisat, CryoSat-2, SARAL/Altika, and Sentinel-3A. We produce a consistent and continuous SLA dataset spanning both ice-free and ice-covered areas of the Southern Ocean (south of 50° S) from 2003 to 2021. This 19-year product provides the longest temporal coverage to date for SLA under sea ice. The resulting SLA fields resolve the large-scale variability and parts of the mesoscale signal, with smooth transitions across the sea-ice edge. We show that the product reliably captures physical signals on timescales longer than 10 days, with estimated uncertainties of 1.3 cm. in the subpolar ocean. Our satellite-based SLA reconstruction compares well with most independent in situ observations from tide gauges and bottom pressure recorders: mean correlation coefficients are 0.58 and 0.66, respectively. The reconstruction fills a key observational gap and offers new opportunities to study trends and interannual variability in sea level and ocean circulation in the Southern Ocean, particularly under sea-ice cover.