Phenological information is vital for dynamic monitoring of crop growth and precision field management. Accurate extraction of phenology benefits a rational analysis of crop’s inter-annual changes in time and space and provides the fundamental data for monitoring of crop growth and crop yield forecasting. Winter wheat planting areas in Hebei, Henan, and Shandong Provinces were taken as study locations. Firstly, the LAI time series in 2015 was smoothed with Savitzky-Golay filtering algorithm, and the day of the maximum value from time series of LAI was taken as heading period. The Savitzky-Golay filtered MODIS LAI was fitted by using the double Logistic function. The day corresponding to the largest second derivative of Logistic LAI curve was considered as the green-up period. Jointing and flowering periods of winter wheat were extracted based on effective accumulated temperature from the existing greenup and heading stages from 2012 to 2014. The method was validated by the observations of phenology at agrometeorological stations in 2015 and the results showed that the phenology agreed well with the observational data of winter wheat. The second derivative method was quite sensitive to mixed pixel and accuracy of Logistic function’ fitting of MODIS LAI. Generally, the extracted greenup stage was delayed, while the extraction of the jointing, heading and flowering stages achieved a high accuracy.