We apply a three-dimensional (3D) full-Stokes model to simulate the evolution of Da Anglong Glacier, a large glacier in the western Tibetan Plateau from the year 2016 to 2098, using projected temperatures and precipitations from the 25-km-resolution RegCM4 nested within three Earth System Models (ESM) simulating the RCP2.6 and RCP8.5 scenarios. The surface mass balance (SMB) is estimated by the degree-day method using a quadratic elevation-dependent precipitation gradient. A geothermal flux of 60 mW m-2 produces a better fit to measured surface velocity than lower heat fluxes and represents a new datum in this region of sparse heat flux observations. The ensemble mean simulated glacier volume loss during 2016–2098 amounts to 38% of the glacier volume in the year 2016 under RCP2.6 and 83% under RCP8.5. Simulation from 2016 to 2098 without ice dynamics leads to an underestimation of ice loss of 22–27% under RCP2.6 and 16–24% under RCP8.5, showing that ice dynamics play an important amplifying factor in ice loss for this glacier, unlike for small Tibetan glaciers where SMB dominates glacier change.