Surface air temperatures affect a diverse set of physical and biological systems in many parts of the world. For regionalscale studies, gridded surface air temperature data sets are frequently used as input variables. Here we evaluate 10 commonly used gridded air temperature products with spatial resolutions ranging from 0.10.1 to 5.05.0, relative to observations from in situ weather stations on the Tibetan Plateau. Gridded temperatures are consistently lower, with mean annual air temperature biases ranging from 4.68C to 1.72C, and root mean square error (RMSE) from 3.24C to 6.11C. The mean biases of mean seasonal air temperatures for spring, summer, autumn, and winter are 3.21C, 2.98C, 2.89C, and 2.98C, respectively, and RMSEs of 4.81C, 4.39C, 4.3C, and 4.7C. Of all 10 products, the China meteorological forcing data set compares best with observations. Annual and season temperatures all increase significantly during 19802018 on the Tibetan Plateau, with winter dominating the warming rate.