Abstract: Taking runoff from the Yingluoxia hydrological station in the upper reaches of Heihe River from 1944 to 2017 as the research object,combining recurence plot theory with chaos theory,the chaotic characteristics of runoff time series were effectively identified under different time scales.The recurence plot and slip window recurrence quantification analysis (RQA) method were used to obtain dynamic information of the original series and extracted the relevant recurence features.Then,drawing the runoff recurence plot under different scales to qualitatively describe runoff chaotic characteristics.Finally,RQA method was used to calculate and quantitatively describe runoff chaotic characteristics.The results showed that runoff chaotic degree varies in different stages,and determinism (DET) of sliding window can describe runoff chaotic degree together with recurence plot.Shannon entropy (ENTR) of sliding window showed that runoff chaotic degree gradually increases with the increase of time scale.And Laminarity (LAM) of sliding window can be tested with coefficient of variation (C_v),and can be used as a new parameter to measure runoff overall dispersion.The recurence plot theory realizes the visualization of runoff chaotic characteristics and is an effective tool to identify the chaotic characteristics of runoff time series.