Abstract. The present study deals with detecting seismic anisotropy parameters beneath southeastern Tibet near Namcha Barwa Mountain using the splitting of direct S waves. We employ the reference station technique to remove the effects of source-side anisotropy. Seismic anisotropy parameters, splitting time delays, and fast polarization directions are estimated through analyses of a total of 501 splitting measurements obtained from direct S waves from 25 earthquakes ( ≥  5.5 magnitude) that were recorded at 42 stations of the Namcha Barwa seismic network. We observe a large variation in time delays ranging from 0.64 to 1.68 s, but in most cases, it is more than 1 s, which suggests a highly anisotropic lithospheric mantle in the region. A comparison between direct S- and SKS-derived splitting parameters shows a close similarity, although some discrepancies exist where null or negligible anisotropy has been reported earlier using SKS. The seismic stations with hitherto null or negligible anisotropy are now supplemented with new measurements with clear anisotropic signatures. Our analyses indicate a sharp change in lateral variations of fast polarization directions (FPDs) from consistent SSW–ENE or W–E to NW–SE direction at the southeastern edge of Tibet. Comparison of the FPDs with Global Positioning System (GPS) measurements, absolute plate motion (APM) directions, and surface geological features indicates that the observed anisotropy and hence inferred deformation patterns are not only due to asthenospheric dynamics but are a combination of lithospheric deformation and sub-lithospheric (asthenospheric) mantle dynamics. Direct S-wave-based station-averaged splitting measurements with increased back-azimuths tend to fill the coverage gaps left in SKS measurements.