With the increased study on the terrestrial carbon cycle,the function of vegetation has become focused on more widely. Net primary productivity (NPP),as one of the characters of vegetation, plays an important role in global change and carbon cycle research in the terrestrial ecosystem. The hilly terrain of southern China, which is the water source of the Pearl River and Yangtze River, has a signification position in China. Therefore, it's crucial to intensify the construction and preservation of the conserving forests at the water sources area of Pearl River Basin. The study on NPP and its tempo-spatial variation in the hilly terrain of southern China would be helpful to understand the growth condition of vegetation and to evaluate the ecological effects of large-scale vegetation construction. In this paper, Carnegie-Ames-Stanford Approach (CASA),a carbon processes-based model based on remotely sensed data, was applied to estimate the terrestrial NPP in hilly terrain of southern China. Monthly MODIS NDVI images,monthly mean temperature,precipitation,solar radiation, vegetation type between 2000 and 2010 were collected for the simulation. The total NPP and mean NPP in hilly terrain of southern China were 117.0 TgC/a and 445.7 gC m~(-2) a~(-1),respectively. The annual NPP ranged from 406.0 to 485.6 gC m~(-2) a~(-1) during 2000 and 2010. However, it had a fluctuant change: It showed a continuous increasing phrase during 2000 and 2004,declined rapidly in 2005 and 2006,increased rapidly in 2007 and 2009,but decreased in 2010. There was an overall increasing trend with annual ratio of 2.28 gC m~(-2) a~(-1). The difference of the spatial pattern of NPP was caused by the changes of land-cover types. The highest average NPP per unit was 501.0 g C m~(-2) a~(-1) in the areas covered by mixed forests, while the lowest average NPP per unit was 390.7 g C m~(-2) a~(-1) in the grasslands. This resulted from the differences in the utilization rates of various resources by various vegetation covers. All of these eight types of vegetation, except for needle-leaved deciduous vegetation, have increased their NPP during 2000 and 2010.1n addition, climate variation is a key factor to influence the change in NPP. The response of NPP to changes in some other factors such as temperature, solar radiation, precipitation and human factors were evaluated and discussed. The change of NPP had notable spatial variability. The NPP had obviously increased in the ecological restoration zone, however it had evidently decreased in the region with rapid urbanization. The change of NPP was the result of interaction of climate change and land-use change. Annual total NPP was significantly positively correlated with annual mean temperature (R~2 = 0.524,P = 0.01),and temperature controlled the annual variation of vegetation growth and prolonged the growth period of vegetation. Precipitation was a control factor of the seasonal change of vegetation growth, but not evidently correlated with annual total NPP. The excessive rainfall, which decreased the solar radiation and diminished the vegetation photosynthesis,was bad for vegetation productivity. The change of land cover was an important factor for vegetation spatial variation. Especially, the implementation of large-scale vegetation constructions by returning farmland to forests or grassland has increased rapidly the forest, leading to a large rise in NPP.