Abstract: Ten-day runoff is affected by many factors, and the variation law is complex. Usually, a single runoff forecasting model cannot reflect the characteristics of incoming water in the basin. Hence, a ten-day runoff fore-casting method based on water and rain information classification is proposed. In this method, the water and rain conditions faced at the time of prediction are divided into six situations, and six different forecasting schemes are proposed for different situations. The prediction effect of eight typical ten-day periods from 2000 to 2010 is tested. The results show that the ten-day runoff forecasting method based on water and rain information classification is effective and can be used to guide the daily operation of Huanren Reservoir.