Abstract: Satellite precipitation products (APHRODITE and TRMM 3B42) and reanalysis data (NCEP) were evaluated quantitatively in this study using rainfall records of gauge-based observation. These datasets were used as forcing data to drive an improved distributed time-variant gain model (DTVGM) to explore whether the precipitation products could be used as a substitute for the rainfall data observed in the Yuanjiang River basin in Yunnan. Results indicate that the accuracy of TRMM is better than the other products. The TRMM datasets of monthly precipitation fit well with the ground observations at catchment scale with correlation coefficients above 0.9 while the correlation of daily precipitation is worse. Streamflow simulations of this study suggest that the simulations of daily streamflow based on the TRMM precipitation products could describe variations in the streamflow, although they have larger deviations than those in the gauge-based observations. Comparison shows that the model performs better after recalibration using precipitation products, suggesting that the precipitation products have a potential to substitute for gauged-based observations.