Sea ice albedo is a critical factor determining the overall energy balance of the ocean-sea ice-atmosphere interface (hereafter referred to as the marine cryosphere) and the seasonal processes of sea ice growth and decay. Small changes in sea ice albedo can affect regional and global climate through the existence of positive feedback loops such as those proposed by Budyko (Tellus, 21, 611-619, 1969). The sensitivity of Arctic regions to various climate change scenarios is largely due to the possibility of positive feedback effects between sea ice and surface albedo. Despite its importance, sea ice albedo has typically been parametrized in models as a simple function of temperature, latitude and ice type. Strong interannual variations in Arctic weather conditions necessitate improved parametrizations of surface albedo. Estimating surface albedo using remote sensing is a promising approach. In this paper we reveal strong relationships between ERS-1 sigma and albedo over multi-year ice. sigma is the average scattering coefficient which expresses the intensity of backscattered microwave energy per unit area received at the Synthetic Aperture Radar (SAR) antenna. The relationship between sigma and albedo is especially strong during the early phases of the melt period. We test general linear models linking observed surface albedo to ERS-1 sigma for two transitional periods: winter to melt conditions; and melt conditions only. We find that change in ERS-1 scattering (Delta sigma ) vs albedo during the melt period produces the optimal empirical model. This model is then used to map sea ice albedo over multi-temporal ERS-1 imagery acquired over the Canadian Arctic Archipelago during the spring of 1995. Classes of multi-year ice albedo are formed based on the confidence limits of the general linear model and are mapped according to the magnitude of Delta sigma in difference images generated from ERS-1 data. First-year ice and rubble ice are segmented from multi-year ice based on their Delta sigma . Firstyear ice albedo is calculated as a linear offset of multi-year ice albedo to account for the contrasting snow distribution over the two ice types. Thematic albedo images indicate that spatial variability in ice types and their respective snow covers is the dominant influence on albedo.