Rough set theory and the fuzzy inference techniques are integrated into the multi factor medium and long term hydrological classification forecast to solve the difficult problem of factors’ choice. The concept of relative classification accuracy and the attributes reduction solution are used to choose the best number of classification and the appropriate forecast factors. The minimal decision solution is regarded as the inference rules to forecast the runoff. With the case of Dahuofang reservoir in China, the influence of the classification and relative accuracy on forecast results are analyzed. The results indicate that the model can effectively solve forecast problem related to complex factors selection. The forecast precision is improved with rough set theory and the forecast result presents not only the level of classification but also the interval value of runoff which can offer abundant reference to make annual plan.
