Typical phenological metrics (start of the season, SOS; end of the season, EOS; and length of the season, LEN) were estimated using moderate resolution imaging spectroradiometer (MODIS)-normalized difference vegetation index (NDVI) data from 2001 to 2016. Then, the phenological interannual variation trend and its spatial pattern were analyzed for the ecological mountain-oasis-desert system (MODS) in Xinjiang and its sub-regions, and the relationship between the vegetation changes and the climate variations were discussed in combination with the meteorological data. The following results were obtained in this study. The phenological characteristics in Xinjiang exhibited obvious zonal and vertical zonal distribution characters, and the altitude plays an important role in the regional differentiation of phenology. The SOS and EOS ranged from mid-March to early-May, and mid-October to late-December, respectively. Contrary to the global phenological change direction, the SOS was delayed in Xinjiang with a delayed amplitude of 1.9 days in every 10 years, the EOS was advanced with an advanced amplitude of 3.66 days in every 10 years, and the shortened LEN amplitude was 5.6 days in every 10 years. The SOS was earlier at the oasis and plain regions but later in the mountain area in the entire Xinjiang area except for the Eastern part. The EOS occurred earlier in the mountain and later in the oasis region of the entire Xinjiang area and different partition zones. The LEN range changed from the oasis and plain to the mountain areas except for the eastern part and showed a vertical zoning character. Based on the redundancy analysis (RDA), which explains most of the relationship information between the phenological characters and the climate factors, the SOS was remarkably influenced by spring temperatures, the year before the winter rainfall and sunshine duration. Summer and autumn precipitation was the important influencing factor for the EOS, which was affected by the temperature and sunshine duration, although the effect was limited. We may have to consider the changing amplitude as the main and secondary factor that affects the phenology metrics when studying the associated interannual variation. This is because various influencing factors may have different effects and change the direction of the advanced or delayed vegetation phenology, which ultimately affects the phenological characteristics of the interannual trend.