Abstract Group velocities for a period range of 6–60 s for the fundamental mode of the Rayleigh wave passing across the Himalaya–Karakoram–Tibet orogen are used to delineate the structure of the upper lithosphere using the data from 35 broadband seismic stations. 2D tomography velocity maps of group velocities were obtained at grids of 1° separation. Redefined local dispersion curves are inverted non-linearly to obtain 1D velocity models and to construct a 3D image of the S-wave structure down to a depth of 90 km. The Moho discontinuity is correlated with c. 4.0 km s −1 S-wave velocity. The results depict a NE-dipping trend of the Moho depth from c. 40 km beneath the frontal part of the Himalaya to up to c. 70–80 km beneath the collision zone before shallowing substantially to c. 40 km beneath the Tarim Basin. The study also reveals thick deposits of sediments in the Indo-Gangetic plains and the Tarim Basin. A broad low-velocity zone at mid-crustal depth in the western Tibetan Plateau, the Karakoram region and the surface-collision part of the India–Eurasia tectonic plates is interpreted as the effect of partial melting and/or the presence of aqueous fluid. The high velocities in the southern deeper part indicate that the lower crust and uppermost mantle of the Indian Plate are dense and cold.