Measurements of shear wave splitting at 43 three-component seismic stationsshow very big difference in anisotropy on both sides of the Indus-Yarlung Zangbo suture(ITS), but little difference on both sides of the older Bangong-Nujiang suture (BNS) and theJinsha River suture (JS) to its north. Obvious discrepancy exists between the anisotropy direc-tion and the superficial tectonic trends, which is not explicable directly by the coherent uppermantle deformation usually supposed to occur in consistency with the trend of a collisional belt.On the other hand, strong spatial relationships are observed from the anisotropy results, such asthe orthogonal directions of anisotropy on both sides of ITS and the good correlation betweenthe region of larger magnitude of anisotropy and the zone of inefficient Sn propagation ofQiangtang as well as the systematic rotation of the directions of anisotropy, which should testifysome much more complicated aspects of the continental convergence mechanism. To the best ofour data, we tend to suppose that the Qinghai-Tibet plateau might result from a mechanismcomplicated by the coexistence of Argand's underthrusting and Dewey's diffuse deformation.