The difficulties involved in manually calibrating the Xinanjiang watershed model have in the past been partly attributable to the lack of robust optimization tools. In fact manual calibration process can be a rather frustrating and time-consuming exercise for an inexperienced person. Therefore,in recent years researchers are exploring ways to incorporate "expert knowledge" of conceptual watershed models into the automatic calibration procedures. This paper presents a brief introduction to the Xinanjiang model and a researdiw Tork on application of SCE-UA method for calibrating Xinanjiang watershed model using hydrological data of three catchments of different size and dimatic mnditioiis. Results show that the overall performance of SCE-UA method for calibrating the Xinanjiang model is very good. On the basis of the results derived from the calibration and verification stages it seems that SCE-UA is capable of finding a global optimum and (onceptually realistic parameter set for the Xinanjiang model.