We have investigated the discrimination power of dual-polarized SAR imagery in ice edge detection. The proposed approach relied largely on the use of HV-polarization imagery consisting of 10 dual-polarized RADARSAT-2 scenes acquired between 20 and 30 January 2010 over the Kara and Barents Seas. The approach utilized the SAR imagery and two SAR based parameters which were transformed using principal component analysis prior to the classification. An exceptional feature in the dataset was the effect of the atmospheric gravity waves on open sea. It was seen that in highly dynamic ice edge conditions the classification approach gave reasonable results if the incidence angle was smaller than 35 degrees, and in some cases also at higher angles. The resulting classification maps were compared to the OSI SAF sea ice edge product which gave consistent results with the SAR data.