Phenology is a critical feature of vegetation, and it could reflect the effects of climate variability and change on vegetation growth. There are complex feedbacks between vegetation phenology and climate change, therefore, researches on the driving mechanism of the change of vegetation phenology would provide a better understanding of the interactions between climate change and phonology. In this study, we selected Qinghai Lake watershed located in the northeastern of the Qinghai-Tibet Plateau as study area and used MODIS 16 days Enhanced Vegetation Index(EVI)composite data to analyze the spatiotemporal pattern and trend of grassland phenology (the start of the growing season, the end of growing season and days of growing season) under the different variations of EVI_max of grassland. The results of the study were concluded as: (1) The changes in EVI_max of grassland showed a diverse trend under the influence of climate change and human activities, an increasing, unchanged and a decreasing trend of EVI_max coexisted at the same time in the whole watershed. (2) Temperature and precipitation increased significantly from 1990 to 2012. Air temperature increased at the rate of 0.42-0.83℃/10a, and precipitation increased at rate of 43.20-44.68 mm/10a. The date of the start of the growing season showed a delay trend around both Gangcha and Tianjun weather station, the date of the end of growing season showed no obvious change but the days of the growing seasons showed a shortening trend between 2001 and 2012. (3) The date of the start of the growing season lasted from April 20 to June 10 and the date of the end of growing season from Mid-August to September 10 in the Qinghai Lake watershed. The places have early dates of the start of the growing season include the eastern and southern lakesides of the Qinghai lake, Buha River estuary, hillslope and flat area of the valley in the western of the watershed. The date of the start of the growing season extended from lakeshore to the high-altitude areas, but vice verse for the date of the end of growing season. (4) Vegetation phenology showed different trend under the different variation of EVI_max. When the EVI_max decreased, the date of the start of the growing season advanced, the date of the end of growing season delayed and the days of the growing season increased. In contrast, the date of the start of the growing season delayed, the date of the end of growing season showed no obvious change and the days of the growing season shorten when theEVI_max increased. The farmland and the places without change in EVI_max had no obvious changes in the date of the start of the growing season and of the end of growing season, however, days of the growing season shorten for farmland. (5) The result suggests that the vegetation phenology change was a complex process, probably affected by multiple factors, such as grassland degradation caused by over-grazing, vegetationEVI_maxincreased under the climate change, enclosed grassland with unreasonable management, the vegetation ecosystem of succession and so on. Therefore, using of remote sensing data for vegetation phenology studies must distinguish dynamic changes of the vegetation of the different experimental samples firstly. Otherwise, it would affect the phenology study results.