Remote sensing of snowpack temperatures from satellites requires knowledge of the spectral emissivity of snow. A model for spectral emissivity is combined with the Planck function to calculate brightness temperature of snow in thermal infrared wavelengths for a range of grain sizes and viewing angles. Emissivity variations caused by density, grain shape, liquid water, and grain size are apparently unimportant, but emissivity varies with viewing angle to produce differences between thermodynamic temperature and brightness temperature as large as 3 K at wavelengths 12 to 14 μm, within the major atmospheric infrared window. This difference is also verified by experimental measurements. An equation to convert brightness temperatures to thermodynamic temperatures is presented, and this is also combined with a dual‐wavelength atmospheric correction method. The spectral emissivity model is also used to calculate an ‘all‐wave’ emissivity of snow: 0.985–0.990 for all grain sizes. Copyright 1982 by the American Geophysical Union.