GLDAS data was used to downscale the water storage data of GRACE gravity satellites, and the actual monthly evapotranspiration (AET) data of 25 sub-basins in the Jinsha River Basin from 2007 to 2016 were reconstructed with the water balance equation. The uncertainty of each water balance parameter was quantified by the Latin hypercube sampling method, and the uncertainty analysis of every parameter and AET was carried out. The results showed that the sampling results were satisfactory when the sample size was set to 415.Precipitation had the largest uncertainty range, followed by the change of water storage, and runoff depth had the smallest uncertainty range. The spatial distribution of AET uncertainty in the Jinsha River Basin showed an increasing trend from west to east and from north to south, with the data ranging from 46.02 mm to 146.19 mm, and it was similar to the spatial distribution of AET signal-to-noise ratio. At the annual and seasonal scales, the AET uncertainty of different land-use types in the river basin was ranked from high to low as follows:woodland, shrub, farmland, grassland, and other land types, and the AET uncertainty varied seasonally, with the largest variation range in summer and the smallest in winter.