In polar regions, interactions between atmosphere and ocean are strongly influenced by the presence/absence of the ice cover and by its thickness, hindering sensible heat fluxes and preventing latent heat loss. Since satellite passive microwave observations became available in the 1970s, significant progress has been made in the study of snow depth and sea-ice concentration and extent in these regions. Estimating sea-ice thickness from satellite data, instead, turned out to be really more difficult. In this work, a new algorithm (named IST) was implemented to estimate sea-ice and snow thickness in Antarctica from SSM/I brightness temperatures using two main ice classification parameters based on microwave polarization and spectral gradient information. Open water, snow presence and thin ice are recognized using a sequence of combined thresholds. All the parameters have been normalized according to PCA analyses responses. Sea-ice results were validated with in situ sea-ice thickness data coming from NIC-NCDC campaigns (in the year 1994) and from Aspect project campaigns (1992–2004). Both showed very good correlations (~ 0.91/~ 0.75) with IST estimates. Then, we could use IST to analyse sea-ice thickness temporal and spatial evolution in the Ross and Weddell Seas during the whole period 1992–2008. The two regions showed a similar seasonal evolution but a very different pluriannual one. No significant negative or positive trends appear from our analyses.