Abstract: The Land use/cover data, soil data, topography data, meteorological data and hydrological data from 2013 to 2017 were used to establish a distributed hydrological model-SWAT (Soil and Water Assessment Tool) in the upper reaches of the Minjiang River. Based on the sensitivity analysis of model parameters, the SWAT model was calibrated and validated to get accurately simulated value of daily runoff which was accessible for hydrological process simulation in upper reaches of Minjiang River. Moreover, with the help of SWAT model, the CI index regarded as an effective index of agricultural drought was calculated to analyze characteristics of the spatiotemporal distribution and variation of drought hazard in this area. Results showed that: (1) SWAT model achieved a good return of applicability analysis in upper reaches of Minjiang River, the Nash-Sutcliffe efficiency coefficient E_NS and the coefficient of determination R²were respectively above 0.7 in both validation and calibration period. (2) SWAT model described the hydrological process well in this area, and the simulated value of daily runoff showed a similar trend of change as the observed value, especially in the main flood season (from June to July). (3) The CI index reflected the spatiotemporal distribution and variation of drought hazard objectively in typical drought year. It indicated that the drought evaluation based on SWAT model was a helpful tool for drought monitoring and early warning at watershed scales.