【Objective】 This study explored the identification method for variation types of river runoff in Yulin area of northern Shaanxi to provide references for further understanding of the change composition, evolution and development mechanism of the river runoff in the Loess Plateau.【Method】 The measured annual runoff sequences at 22 hydrological stations in Yulin, Shaanxi were collected. The empirical mode decomposition(EMD) was combined with the second derivative, STARS and iterative cumulative sum of squares(ICSS) to decompose the streamflow series into intrinsic mode functions (IMFs,i· e. the high frequency component) and residual component (i· e· the low frequency component). The second derivative,STARS and ICSS methods were used to detect the trend changes and the abrupt changes in mean and variance in the runoff sequences for effective identification of compound variation types and characteristics. 【Result】 The annual runoff sequences of 22 hydrological stations in Yulin, Shaanxi showed attenuation trend,and the annual runoff attenuation rates were significantly different (-0. 06%--54. 73%),which posed a serious threat to regional water security in Yulin. The regional variation types of the annual runoff sequence were complex and variable,there were 8 types including no variation,trend variation,mean variation, variance variation, double variation in trend and mean,double variation in trend and variance, double variation in mean and variance,and triple variation in trend,mean and variance The trend variations with rising before falling and with falling before rising mainly occurred around 1960 and 2000,respectively. Mean variation with jumping down was at around 1970 and variance variation occurred in 1960s and 1970s.【Conclusion】 The variation degrees of runoff time sequences in different drainage basins of Yulin in northern Shaanxi were different and the types were complex. The trend variation,mean variation and variance variation were in balanced situation,and the compound type was gradually dominant The changes of river streamflow series in Yulin region are being more complicated by the influences of climate change, underlying surface changes and human activities.