Land use and land cover change (LULCC) can influence regional climate by altering the surface roughness, soil moisture, and heat flux partition. LULCC information therefore is important for providing a better understanding of land-atmosphere interaction. A key requirement for such data for use in climate studies is the generation of land use maps on at least an annual basis. However, current continuous annual land use and land cover maps are only available from 2001 to present, limiting the span of the period that can be studied. This study applied a random forests classifier based on nineteen phenological metrics to produce land use and land cover maps of China from 1982 to 2012 using Advanced Very High Resolution Radiometer (AVHRR) third generation NDVI (NDVI3g) dataset, and Moderate-resolution Imaging Spectroradiometer (MODIS) land cover type product (MCD12Q1). The overall accuracy is 86.3%, which indicates the reliability of the maps.