Abstract: Hilbert-Huang transform (HHT) is a new approach to analyze nonlinear, non-stationary time series. This paper presents an application of this transform to the annual runoff at the Qingyangcha hydrological station of the Dalihe River basin during 1959-2000. Using HHT, we calculated the intrinsic mode function (IMF) component and trend term Res of the response period and its HHT spectra. The results show that the runoff displays a significant downward trend with a downward rate of roughly 436980 m3/a and it takes variation periods of 29a, 15a, 6a, and 3a. Then, they were verified by using the methods of Mann-Kendall (MK) test and Mexhat wavelet transform. This shows that the HHT results agree with those by MK test and wavelet transform while HHT has a better time resolution than wavelet transform. And the MK tests detected a mutation point of the runoff series that had occurred in 1971. All the results indicate that HHT is an effective approach to nonlinear, non-stationary hydrological series and that using different methods for analysis can mine more useful information from the series and better understand hydrological processes.