By using conventional observations of snow depth, snow day, satellite remote of sensing snow depth, snow cover fraction, and snow water equivalent data in winter over China for 1979-2006, comparisons on their spatial and time variations were conducted. It is determined that the variations in the five types of snow datasets are out of phase between Xinjiang Uygur Zizhiqu, Northeast-Inner Mongolia, and the Tibetan Plateau, South China, in the leading mode of empirical orthogonal functions. Namely, the positive (negative) anomalies in snow over Xinjiang Uygur Zizhiqu and Northeast-Inner Mongolia are associated with negative (positive) anomalies over the Tibetan Plateau and South of the Yangtze River valley. Over North China, snow depths, snow cover fractions, and snow days increase with time whereas snow water equivalents reduce with time. The variations in snow cover fraction and snow day over the Tibetan Plateau are opposite those in North China. Over South China, decreases in snow depths and snow days were observed. In general, satellite remote of sensing snow data are suitable for plateaus and mountains where fewer observations are available and for large-scale snow research over the Northern Hemisphere; observation snow data are suitable for Central-West China or the plain areas.