Snow can act as an effective insulator, protecting the soil from cold winter atmospheric temperatures. The timing of first snow fall, and its thickness during the coldest months of the winter and the structure of the snowpack are important factors in determining the soil temperature drop over the winter. Although snow cover is often assesed late in the season, prior to melt, daily observations yield a basis for analyzing the effect of snowfall timing on soil temperature. Multiyear records are required to compare the effect of snowpack thickness.Surprisingly few observational records of this type are in the public domain for the Arctic. We present snow depths and soil temperature observation records up to 7 years in length and include sites on the North Slope of Alaska, at Ny Alesund, Spitsbergen and on Samoylov Island in the Lena Delta, Siberia. Vertical arrays of sensors were installed in and below the seasonally thawing layer of soil, from near the ground surface into the upper permafrost. At each site, the soil profile was characterized, and the depths and bulk densities of the soil horizons were measured. Snow depth was measured using either a snow probe, snow pit excavation or an automatically logged ultrasonic snow depth sensor. The observations cover a wide range of snow depths and soil temperatures: maximum annual snow depth at all sites ranged from less than 0.2 m to over 1.5 m in depth; the earliest soil freezing following snow fall (below 0.1 m) began on August 30 and some soil horizons unremained unfrozen until mid-December. The effect of an intermediate snow cover, under normal winter conditions, can be to delay freeze-back by at least 1 month at most depths. The combined effects of the exposed frost boil mineral soil, shallow snow depth and cold winter temperatures lead to extremely cold soil temperatures in the soil.