Abstract: Based on the monthly flow series of the two gauging stations, Zagunao and Sangping on Zagunao River, upper stream of the Minjiang, the wavelet transform was adopted to analyze the flow trend and periodic variation of the Zagunao Watershed. The standardized time series were decomposed by Multi-Resolution Analysis (MRA) using the db3 wavelet function. The reconstruction of the lowest frequency part shows the descending trends for the two flow series and the trend descends more rapidly in regions with worse vegetation conditions than those with better vegetation conditions, which testifies the equalizing effect of forests on stream flow. Besides the vegetation, human activities such as increasing irrigation and hydroelectric power stations are also important factors on the descending flow. Finally, continuous wavelet transform (CWT) was carried out to identify the periodic variation of the two standardized series using the complex valued Morlet function. And the wavelet variance is employed to identify the dominant period, which results in 6 years for Zagunao and 7-8 years for Sangping approximately. The results indicate that the wavelet transform is an effective tool for nonstationary stochastic series analysis, which has great potential in hydrology and water resources research.