Potential evapotranspiration is an effective tool for monitoring energy balance and humidity transfer, and can serve as an important indicator of whether extreme meteorological events occur. The Yarlung Zangbo River basin is extremely sensitive and vulnerable to climate change, with complex terrain conditions, large geographical span, sparse observation stations in the basin, great difficulty in ground observation, and shortage of data, making it difficult to obtain high-precision and large-scale potential evapotranspiration data. At present, Penman Monteith formula is mainly used to estimate the potential evapotranspiration in the the Yarlung Zangbo River basin, and there are relatively few studies on the evaluation and fusion of potential evapotranspiration remote sensing products. Therefore, with the Yarlung Zangbo River basin as the research area, the multi-source remote sensing datas are used to fuse a set of potential evapotranspiration fusion data with better adaptability, in order to reveal the potential evapotranspiration changes in the Yarlung Zangbo River basin, and provide a scientific basis for the study of water and heat cycle and ecological hydrological processes of the basin under global climate change.Based on the evaporation pan datas observed by the the Yarlung Zangbo River basin meteorological station from 2001 to 2018, Pearson correlation coefficient, bias and root mean square error were selected to analyze and evaluate four commonly used remote sensing potential evapotranspiration products GLEAM, MOD16A2, GLDAS_Noah and ERA5. Then, a better adaptive potential evapotranspiration fusion data was generated by Triple Collocation (TC) method and least square method, and the spatiotemporal distribution of potential evapotranspiration in the study area was discussed. At site scale and basin scale, ERA5 potential evapotranspiration products had good adaptability in the Yarlung Zangbo River, and its assessment accuracy was significantly better than GLEAM, MOD16A2 and GLDAS_Noah potential evapotranspiration products. In terms of spatial distribution of station accuracy, except GLEAM, the correlation of the other three remote sensing data was most significant in the middle reaches of the basin. The relative deviation of MOD16A2 and ERA5 was small, and the spatial distribution of their root mean square error of the two was opposite to the relative deviation. The weights of MOD16A2, GLDAS_Noah and ERA5 remote sensing datas in the fused data were 31.12%, 30.64% and 38.24%, respectively. The uncertainty analysis of the three remote sensing products showed that the spatial distribution of errors in the upper reaches of the Yarlung Zangbo River basin was gradually decreasing from north to south, but the spatial distribution of errors in other parts was different. The spatial distribution of weights was basically opposite to the spatial distribution of errors. The four types of data had large multi-year mean potential evapotranspiration in the middle of the Yarlung Zangbo River basin, and the difference was whether there was a greater potential evapotranspiration in the southeast corner of the basin. The multi-year average potential evapotranspiration was in the order: ERA5