Abstract The authors use a statistical regional climate model [Statistical Regional Model (STAR)] to project the Tibetan Plateau (TP) climate for the period 2015–50. Reanalysis datasets covering 1958–2001 are used as a substitute of observations and resampled by STAR to optimally fit prescribed linear temperature trends derived from the Max Planck Institute Earth System Model (MPI-ESM) simulations for phase 5 of the Coupled Model Intercomparison Project (CMIP5) under the representative concentration pathway 2.6 (RCP2.6) and RCP4.5 scenarios. To assess the related uncertainty, temperature trends from carefully selected best/worst ensemble members are considered. In addition, an extra projection is forced by observed temperature trends in 1958–2001. The following results are obtained: (i) Spatial average temperature will increase by 0.6°–0.9°C; the increase exceeds 1°C in all months except in boreal summer, thus indicating a reduced annual cycle; and daily minimum temperature rises faster than daily maximum temperature, resulting in a narrowing of the diurnal range of near-surface temperature. (ii) Precipitation increase mainly occurs in early summer and autumn possibly because of an earlier onset and later withdrawal of the Asian summer monsoon. (iii) Both frost and ice days decrease by 1–2 days in spring, early summer, and autumn, and the decrease of frost days on the annual course is inversely related to the precipitation increase. (iv) Degree-days increase all over the TP with peak amplitude in the Qaidam Basin and the southern TP periphery, which will result in distinct melting of the local seasonal frozen ground, and the annual temperature range will decrease with stronger amplitude in south TP.