Historical simulation outputs of climate models from the Coupled Model Inter-comparison Program Phase 6 (CMIP6) together with the GlobSnow monthly snow water equivalent (SWE) products were used to evaluate the model performance in simulating the Eurasian SWE of winter. Most models can relatively well reproduce the basic features of the climatological Eurasian winter SWE. There are great differences in different models for the trend of SWE, but the multi-model ensemble (MME) can improve the simulation ability. The spatial and temporal characteristics of winter SWE from CMIP6 model simulations and observations were analyzed using the Empirical Orthogonal Function (EOF) analysis, and the results suggested that only a small number of CMIP6 models could reproduce main features of the first eigenvector, but MME can improve the performance. Furthermore, MME can also reproduce the response of Eurasian SWE to precipitation and surface air temperature during winter. The projection of Eurasian winter SWE in the 21st century was estimated by using the CMIP6 MME results under different emission scenarios of Shared socioeconomic pathways (SSPs). With respect to the reference period 1995?2014, projections of SWE by the MME under four scenarios (SSP126, ssp245, ssp370 and ssp585) all shown increasing trend in the northeastern Eurasia continent and decreasing trend in continental Europe to the west of 90°E. There was little difference in SWE change under four SSPs in the early 21st ^Century. But the difference became larger in the late 21st ^{century. The amplitudes of SWE changes} would become larger with time. Further analysis shows that the higher temperature and increased precipitation were conductive to the SWE increase in the northeastern Eurasia.