Аннотация:We have combined continuous and survey-mode GPS observations from eastern Russia and Japan with those of the global IGS network between 1995 and 1999. The rms deviations of horizontal velocities for 16 stations encompassing all of northern Eurasia from Svalbard Island to Siberia is less than 1 mm/yr. We used these stations to define a Eurasia-fixed frame of reference and found that stations in eastern Asia (Irkutsk, Vladivostok, Shanghai, Petropavlovsk, Yuzhno-Sakhalinsk, Taejon, Tsukuba) move 4-6 mm/yr less eastward than in recently published solutions in which velocities in the ITRF96 no-net-rotation (NNR) frame are rotated to a Eurasian frame using the NUVEL-1A plate model. We believe that this difference results from a different realization of the NNR frame in NUVEL-1A and ITRF96 as suggested by Zhang et al. [Geophys. Res. Lett., 26, 2813, 1999].With the frame defined using a set of stations spanning all of stable Eurasia, the velocities of Vladivostok (VLAD) and Yuzhno-Sakhalinsk (YSSK) are less than 4 mm/yr with respect to Eurasia. Whether the area between these stations and Lake Baikal constitutes a rigid Amurian block or represents a diffuse boundary between South China and Eurasia cannot be determined from the current, sparse set of geodetic measurements.The velocity at Bangalore (38 $\pm$2) implies that the northward motion of India is 20\% slower than that predicted by NUVEL-1A. The uncertainty in the Euler vector defining the relative rotation of Eurasia and North America is a factor of two smaller than previous geodetic estimates and is insensitive to whether we define Eurasia using the full 16 stations or using subsets encompassing only the eastern or western parts of the plate. Our pole is 1200 km northwest of the NUVEL-1A pole, implying convergence across the southern Cherskiiy Ranges of northeast Russia.