Freshly fallen snow, drifting snow, and surface snow samples were collected between February 1997 and January 1998 at the Dome Fuji Station (77.19¿S, 39.42¿E, 3,810 meters above sea level (MASL)) located approximately 1000 km from the nearest coast. All the collected samples were analyzed for δ18O at the National Institute of Polar Research (NIPR) at Tokyo. The temperature at the top of the boundary layer (TBL) was measured at least twice a week. As expected, all collected samples show seasonal variations with higher values in summer and lower values in winter. However, the relationship with the surface air temperature (Ta) was found to be different for each species of samples. During the sampling year, the surface air temperature, which was measured using the classical methods, was found to be about 3¿C higher than the mean surface temperature Ts, defined as the firn layer temperature at 10 m depth. One important process that leads to differences between the freshly deposited snow and the surface firn layer is the sublimation of air moisture during the winter months. Remarkable progress has recently been made to explain the δ2H and δ18O distribution patterns for Polar Regions with GCMs. Detailed observations at the Dome Fuji station reveal that δ18O values for the individual snowfalls are only partly explained by the improved Rayleigh type models, and that the processes after snow deposition also have to be taken into account.