Vegetation phenology, a regular and periodic phenomenon in nature, is an important indicator for natural environment, especially climate change. The study of spatiotemporal variations of vegetation phenology is of great significance for monitoring the changes of terrestrial vegetation. In this study, the Savitzky-Golay (S-G) filtering method was used to reconstruct time-series MODIS enhanced vegetation index (EVI) data in the Qinling Mountains from 2001 to 2018. The dynamic threshold method was used to extract the spring phenological parameter (start of growth season, SOS). The correlation between multi-year mean SOS and interannual variation with altitude and slope was analyzed. The results showed that SOS was delayed by 1.82 d with every 100 m increase in altitude in the Qinling Mountains. The interannual change trends of SOS mainly concentrated in 0-5 d·(10 a)-1. The pixels with delaying trend were mainly distributed in low-altitude regions, with the delaying degree being gradually decreased with the elevation. The interannual change trend of SOS in high-altitude regions was more complex than that in lower-altitude regions. The multi-year average SOS in the northern slope was approximately 2.9 d earlier than that of the southern slope, whereas the southern slope had a more significant advancing trend. The interannual change trends of SOS in both north and south slopes showed a delaying trend in low-altitude, with little difference between north and south slopes. The advancing trend in middle and high altitude was significantly different.