This is a collection of shallow snow density data sets from research station Neumayer II, Neumayer III and the Kottas traverse between Neumayer, Kottas camp and Kohnen station. The original snow density data are available as Excel and txt files and as density-depth profiles in png format. The data were collected between 2001 and 2021. Measurements in 2019, (2020) and 2021 were taken on a monthly basis. Before this period, sampling took place on a rather irregular basis, but often once a year, at Neumayer station in most cases by members of the overwintering team, lead by the overwinterer for the air chemistry observatory (SPUSO). Sampling during the Kottas traverse was carried out by members of the traverse personnel. The sampling devices used were • conventional steal cylinders with a length of about 20 cm and an inner diameter of 6 cm and a wall thickness of 1 mm; the exact measurements were recorded in the protocols. Due to damage by ice layers the cylinders often have to be sharpened again and therefore become shorter over time ( (method listed in sensor.awi.de as awi_snow_sampler_3). • single-tube carbon fibre liners, as described in Schaller et al. (2016) with a length of 1 m and an inner diameter of around 10 cm and a wall thickness of 1 mm (method listed in sensor.awi.de as awi_snow_sampler_2); • dual-tube carbon liner probing system developed by Dallmayr et al. (2020) (method listed in sensor.awi.de as awi_snow_sampler_1). More details about sampling event nomenclature is provided below in the file description. For the steal cylinder sampling method, sample density was determined for individual depth sections, thus providing a stratigraphic density profile. For both liner methods, density was determined as the bulk density, i.e. the average density of the whole snow core sampled with the carbon liner tube. If it was possible to remove the snow core in a mostly intact way from the liner, it was cut into samples in a aluminium tray, as described in Schaller et al. (2016). This also eventually provides a density profile. Sample depth vary, as thick ice layers often inhibit deeper penetration with either method. In some cases, the carbon liner was damaged, resulting in a non-cylindrical form of the snow sample. Such events were noted in the protocols and an attempt made to correct the density accordingly. A description of the density data sets' is provided in the MSc thesis by Cedric Hecht under supervision of Olaf Eisen.