In this study, we used NDVI time-series datasets to investigate the variations in the starting dates of first growing season (SGS) in China′s cropland over the past 20 years. To do so, a smoothing algorithm based on an asymmetric Gaussian function was first performed on the NDVI time-series data to minimize the effects of anomalous values caused by atmospheric haze and cloud contamination. The SGS of cropland was then estimated from three smoothed NVDI time-series dataset for early 1980s, 1990s and 2000s, respectively. The resulting three datasets were overlaid together to calculate the SGS variations for the period from early 1980s to early 1990s and for the period from early 1990s to early 2000s. The results show that the SGS in China′s cropland becomes progressively later as moving northward from the South of China, in good accordance with the temperature and precipitation gradients in China. The SGS of China′s cropland varies considerably with location and investigation period, and its general trend is characterized by a significant advance in these two periods. This study supported the conclusions from previous studies that the SGS has advanced over the past two decades at the middle and high latitudes in the northern hemisphere. Several factors can profoundly influence cropland phenological status, such as climate change, soils and human activities. How to discriminate the impacts of biophysical forces and anthropogenic drivers on phenological events of cultivation remains a great challenge for further studies.