Since International Association of Hydrological Science (IAHS) initiated the prediction in ungauged basin (PUB) programe, the regionalization has become a common method for identifying hydrological model parameters in ungauged basins. However, some problems exist in the regionalization method, which was commonly used for parameter identification of hydrological model in ungauged basins, such as different relationships between rainfall and runoff in similar basin, unobvious correlation between hydrological model parameters and basin characteristics, and difficult to select a reference basin in wide range of data-deficient areas, etc. It is significant to study the method of calibrating hydrological model by using the data outside the stream flow. A method was presented to calibrate the parameters of grid-based Sacramento (GSAC) model by using evapotranspiration (ET) data from global land data assimilation system (GLDAS). Firstly, a spatiotemporal-grid corresponding relationship between GLDAS ET and ET simulated by GSAC model was established. Secondly, a evaluation index based on Nash-Sutcliffe efficiency coefficient was constructed to measure the fitting effect between GLDAS ET and ET simulated by GSAC model on 0.25° grid. Finally, validation of GSAC model was carried out based on the GLDAS ET. The results of application in the Hulan River Basin indicated that GLDAS ET can better simulate the actual ET of the Hulan River Basin so that provided a useful input data for calibrating parameters of GSAC;the runoff Nash-Sutcliffe efficiency coefficient of GSAC model calibrated by GLDAS ET were 0.81 and 0.77 in the calibration and validation periods, respectively, and the results were similar to the corresponding values of GSAC model calibrated by runoff data.