Abstract: In order to meet the needs of flood control and water storage service in hydrometeorological forecast, the periodic features of the first and last heavy rainfall in the Jialing River of upper Yangtze River are analyzed, by applying the power spectrum method to daily precipitation data at meteorological stations. A prediction method for the fi rst and last heavy rainfall are studied by means of stepwise regression applied to climate system monitoring indexes provided by China Meteorological Administration. The results indicate that there exist obvious periodic features on the first and last heavy rainfall in Jialing River.The regression equations are set up with monitoring indexes. All multiple correlation coef ficients of every equations have passed 0.01% signi ficant test level. This means that the equations fi t observations very well. The independent sample test results for 2006-2017 from the regression equations show that anomaly sign consistent rates are both 66.7%. Those results mean that this method has a certain reference value.