The trends and variations in precipitable water (PW) over the Tibetan Plateau (TP) from 2000 to 2010 are analysed using the monthly mean PW datasets from Moderate Resolution Imaging Spectroradiometer (MODIS) and ERA-Interim reanalysis of European Centre for Medium-Range Weather Forecasts (ECMWF). Three different methods (including linear fit, non-parametric and non-linear) are used to estimate PW trends and their significance. The trend patterns derived from MODIS and ECMWF are very similar except for the magnitude. Increasing trends are found in the eastern and western TP; with decreasing trends in the middle. Atmospheric water vapour is more changeable compared to the surrounding areas and large PW anomalies occur frequently in July and August over the TP. Regression analysis between monthly mean anomalies of ECMWF PW and ECMWF surface air temperature (SAT) indicates reanalyses data show larger regression slope (7.36 ± 0.75 % K–1) over the TP than the MODIS observations (4.81 ± 0.72 % K–1) mainly in the non-monsoon season. The comparison of the vertically integrated moisture flux and its divergence between the wettest and driest monsoon seasons indicates that the PW spatial variability in monsoon seasons over the TP is affected by the large-scale atmospheric circulation. Although the TP region is under the influence of one large-scale circulation pattern (the monsoon) and the consistent warming background, the PW trends and the relation of PW and temperature are spatially heterogeneous, which is particularly important for predicting hydrological changes in this region.