A large amount of hydrological data stores in hydrological institutions in our country. In order to improve the accuracy of hydrological forecasts and to lower the complexity of the forecast, a important research direction is how to use intelligent algorithm to analyze the data. T-S-K fuzzy logic system is one of the opinions. The Takagi- Sugeno-Kang (T-S-K) fuzzy system is adopted in this study to establish the relationship between the observed downstream discharges and the observed upstream discharges in the Fu River basin. The result shows that the quality of the modle built by T -S -K fuzzy logic system is good, the forecast accuracy is high, the forecast model quite meet the practical application requirements.