It 's of significant importance to understand the effect of LUCC on vegetation net primary productivity (NPP) in global carbon cycle and food security. This study aimed to investigate how the land cover change caused by urban expansion affects the regional NPP in Guangdong Province. The vegetation type data, land cover data, moderate-resolution imaging spectroradiometer (MODIS) normalized difference vegetation index (NDVI) and meteorological data were combined to simulate the spatial distribution of NPP in the year 2000, 2005 and 2010 in Guangdong Province by using the improved Carnegie-Ames-Stanford Approach (CASA) model. Guangdong Province is divided into four ecological regions: Nanling region, Yuezhong region, the Pearl River Delta and Yuenan region. Then, the spatial pattern of NPP and spatio-temporal dynamics of land cover change were analyzed, and the regional differences of NPP gains/losses caused by land cover change were compared. The results show that: 1) The NPP is heterogeneous in space and forms a series of zones with high value and low value; the Yuezhong region has the highest annual mean NPP, and the lowest NPP value is in the Pearl River Delta. 2) The change of land cover in Guangdong Province increased during 2000-2010, with the maximum changes happening in urban land and cropland, and the Pearl River Delta and Yuezhong region had the most change. 3) Under the condition of climate change, the land cover change resulting from urban expansion caused the overall NPP loss in Guangdong Province; the primary loss is caused by the conversion from croplands and forests to land covers with low productivity, especially the urban land. The NPP losses in different ecological regions are greatly different, the Pearl River Delta and Yuezhong region having more serious NPP loss during 2000-2005, and with a decrease of NPP loss in the Pearl River Delta and increases in other ecological regions during 2005-2010, which indicates that the effects of urban expansion on NPP are complex.