Remotely sensed derivation of sea-ice thickness requires sea·ice density. Sea-ice density was estimated with three techniques during the second Sea Ice Physics and Ecosystem eXperimett (SIPEX-II, September-November 2012, East Antarctica). The sea ice was first-year highly deformed, mean thicknsss 1.2 m with layers, consistent with rafting, and 6-7/10 columnar ice and 3/10 granular ice. Ice density was found to be lower than values (900-920 kg m−3 used previously to derive ice thickness,, with columnar ice mean density of 870 kg m− 3. At two different ice stations the mean density of the ice was 800 kg m–3, the lower density reflecting a high percentage of porous granular ice at the second station. Error estimates for mass/volume and liquid/solid water methods are presented. With 0.1 m long, 0.1 m core samples, the error on individual density estimates is 28 kg m-3. Errors are larger for smaller machined blocks. Errors increase to 46 kg m-3 if the liquid/solid volume method is used. The mass/vouume method has a low bias due to brine drainage of at least 5%. Bulk densities estimated from ice and snow measurements along 100 m transects were high, and likely unrealistic as the assumption of isostatcc balance is not suitable over these length scales in deformed ice.