This　paper　presents　a　study　on　an　ancient　river-damming　landslide　in　the　SE　Tibet　Plateau,　China,　with　a　focus　on　time-dependent　gravitational　creep　leading　to　slope　failure　associated　with　progressive　fragmentation　during　motion.　Field　investigation　shows　that　the　landslide,　with　an　estimated　volume　of　4.9　x　10(7)　m(3),　is　a　translational　toe　buckling　slide.　Outcrops　of　landslide　deposits,　buckling,　toe　shear,　residual　landslide　dam,　and　lacustrine　sediments　are　distributed　at　the　slope　base.　The　landslide　deposits　formed　a　landslide　dam　over　60　m　high　and　at　one　time　blocked　the　Jinsha　River.　Optically　stimulated　luminescence　dating　for　the　lacustrine　sediments　indicates　that　the　landslide　occurred　at　least　2,600　years　ago.　To　investigate　the　progressive　evolution　and　failure　behavior　of　the　landslide,　numerical　simulations　using　the　distinct　element　method　are　conducted.　The　results　show　that　the　evolution　of　the　landslide　could　be　divided　into　three　stages:　a　time-dependent　gravitational　creep　process,　rapid　failure,　and　granular　flow　deposition.　It　probably　began　as　a　long-term　gravitationally　induced　buckling　of　amphibolite　rock　slabs　along　a　weak　interlayer　composed　of　mica　schist　which　was　followed　by　progressive　fragmentation　during　flow-like　motion,　evolving　into　a　flow-like　movement,　which　deposited　sediments　in　the　river　valley.　According　to　numerical　modeling　results,　the　rapid　failure　stage　lasted　35　s　from　the　onset　of　sudden　failure　to　final　deposition,　with　an　estimated　maximum　movement　rate　of　26.8　m/s.　The　simulated　topography　is　close　to　the　post-landslide　topography.　Based　on　field　investigation　and　numerical　simulation,　it　can　be　found　that　the　mica　schist　interlayer　and　bedding　planes　are　responsible　for　the　slope　instability,　while　strong　toe　erosion　caused　by　the　Jinsha　River　caused　the　layered　rock　mass　to　buckle　intensively.　Rainfall　or　an　earthquake　cannot　be　ruled　out　as　a　potential　trigger　of　the　landslide,　considering　the　climate　condition　and　the　seismic　activity　on　centennial　to　millennial　timescales　in　the　study　area.