Runoff is an important index which can reflect the climate and geographical environment change in a basin, analyzing its multi-scale variability and influencing factors is of great significance for the understanding of hydrological changes and the protection of water resources security under the background of climate change. The Xiaojue River Sub-basin(XJ) and Ye River Sub-basin (YR) in the mountainous Hutuo River Basin were selected to study the multi-scale runoff variability of tendencies, abrupt changes and periods using M-K trend test, abrupt change test and Morlet wavelet analysis based on the monitored hydrological and meteorological data. Meanwhile, the relative importance of driving forces (temperature, precipitation, potential evapotranspiration and leaf area index) at different periods (flood season and non-flood season) on monthly runoff changes were revealed by correlation analysis and random forest model. Results showed that from 1960 to 2018, the annual average runoff depth of the two sub-basins showed a significantly decreasing trend, but the mean annual runoff depth of the YR (69.34 mm) was greater than that of the XJ (39.07 mm), which decreased at a rate of 13.0 mm/10a and 8.7 mm/10a, respectively; the monthly and seasonal runoff depth also decreased significantly, especially in summer. An abrupt change was observed in 1981 and 1979 in the XJ and YR, respectively, after the abrupt change the decrease rate of the runoff depth slowed down, and the frequency of extreme runoff decreased, with the largest decline of runoff spotted in August. The main periods of runoff depth were 15, 8 , 5 a in the XJ and 16 , 10, 5 a in the YR. Additionally, there were five periods of dry and wet at the timescale of the first principal period. The results of the correlation analysis and random forest model indicated that precipitation was the dominant factor affecting the long-term runoff changes, especially in the flood season, whereas leaf area index was another important factor affecting the runoff changes in the non-flood season.