The knowledge of the thermal condition within the active layer is of great significance for understanding the variation characteristics of the active layer,the thermal effect of the upper underlying ground and for predicting climate change on the Qinghai-Tibet Plateau. In this paper,the GIPL2 model was used to simulate the soil thermal variations at different depths under different vegetation in permafrost regions over the Qinghai-Tibet Plateau. The simulated results indicate that the GIPL2 model has performed better at the alpine steppe observation site (QT06),but at the alpine swamp meadow observation site (QT03) it has performed bad; between both there are the alpine meadow observation site (QT01),the alpine desert steppe observation site (QT05) and the alpine steppe observation site (QT04) . The root-mean-square errors at QT01,QT03,QT04,QT05 and QT06 are 0. 67,1. 29,0. 73,0. 7 and 0. 56 ℃,respectively; their correlation coefficients are 0. 99,0. 87,0. 98 , 0. 98 and 0. 96 and their mean bias errors are 0. 37,0. 61,0. 31,0. 45 and 0. 16 ℃,respectively. The GIPL2 model has better result at QT06,which probably results from the soil stratification used in the model is closer to real condition. The simulated results at QT03 is worse,which may be due to the existence of gravel in the soil and the thermal conductivity parameter in the model has not considered the gravel content. Overall,the GIPL2 model can simulate soil temperature of multiple soil layers in permafrost regions on the plateau. It means that the GIPL2 model could be used to simulate the thermal regime of the active layer of the permafrost on the Qinghai-Tibet Plateau.