Abstract: Phenology is the formed growth and development law of vegetation to adapt to the long-term cyclical changes of environment. It has great value to research crop phenology for predicting the effects of agriculture, crop yield and climate change. At the same time, phenology is also an important parameter of land surface process model and global vegetation model. [Objective] Compared with the traditional site observation method, remote sensing can overcome the shortcomings of traditional fixed point observation, and is widely used in this field. The current remote sensing monitoring method focuses on the morphological feature points of the time series curve and its corresponding phenological information. The relative relationship between phonological information and morphological features is rarely considered in the research process. Aiming at the problem of existing researches, this paper improved the remote sensing extraction method of phenological information, and extracted the crop phenology information by using the MODIS NDVI time series data from 2006 to 2015 in the main agricultural areas of Yiluo River Basin. [Method] First, the NDVI data from MODIS MOD13Q1 and MYD13Q1 products were combined to form 8-day interval time series data, increasing the temporal resolution of data, and then Savitzky-Golay filter was used to remove noise information and form a smooth curve because the local fitting is more suitable to show complex phenological phenomenon. After that, the phenological dates were defined by the relative position to the key morphological feature points according to the biological physiological characteristics, combined with the characteristics of remote sensing data. They are emergence date，dormancy date，green-up date，heading date and maturity date respectively for winter wheat, and emergence date，jointing data, tasseling data and maturity date respectively for summer maize. Finally, the key phenological information of winter wheat and summer maize were extracted from the processed time series curve based on the definition of each phenology in the study area. [Result] The results show that the relative variation of emergence date is the smallest, while the heading stage of winter wheat and the tasseling data of summer maize are the most. For winter wheat and summer maize, the emergence date is separately concentrated in mid-October and mid-June, while the maturity date is separately concentrated at the end of May and the end of September during the ten years. They are relatively stable because of planting habits. The heading stage of winter wheat is almost distributed throughout April, and the tasseling data of summer maize is almost distributed throughout August. The monitored phonological information is objective and credible compared with previous observations. However, there are clear variations for each phenology between the years when crops are affected by climate change and abnormal weather. Among them, it is found that the maximum variation in the same phenology is about twenty days during the ten years in the study area. [Conclusion] This demonstrates that remote sensing technology can objectively reflect what has happened, and the proposed method is feasible and effective in this paper. It is of great significance to grasp timely phenological information for agricultural production and management, as well as in-depth agricultural remote sensing researches.