The duration and extent of snow cover is expected to change rapidly with climate change. Therefore, there is a need for improved monitoring of snow for the benefit of forecasting, impact assessments and the population at large. Remotely sensed techniques prove useful for remote areas where there are few field-based monitoring stations. This paper reports on a study of snow season using snow cover area fraction data from the two northernmost counties in Norway, Troms and Finnmark. The data are derived from the daily 500 m standard snow product (MOD10A1) from the NASA Terra MODerate Resolution Imaging Spectroradiometer (MODIS) sensor for the 2000–2010 period. This dataset has been processed with multi-temporal interpolation to eliminate clouds. The resulting cloud-free daily time series of snow cover fraction maps, have subsequently been used to derive the first and last snow-free day for the entire study area. In spring, the correlation between the first snow-free day mapped by MODIS data and snow data from 40 meteorological stations was highly significant (p < 0.05) for 36 of the stations, and with a of bias of less than 10 days for 34 of the stations. In autumn, 31 of the stations show highly significant (p < 0.05) correlation with MODIS data, and the bias was less than 10 days for 27 of the stations. However, in some areas and some years, the start and end of the snow season could not be detected due to long overcast periods. In spring 2002 and 2004 the first snow-free day was early, but arrived late in 2000, 2005 and 2008. In autumn 2009 snowfall arrived more than 7 days earlier in 50% of the study area as compared to the 2000–2010 average. MODIS-based snow season products will be applicable for a wide range of sectors including hydrology, nature-based industries, climate change studies and ecology. Therefore refinement and further testing of this method should be encouraged.