Abstract: Vegetation Net Primary Productivity （NPP） is an important index to evaluate the terrestrial ecosystem carbon sink and regulation process， however， the estimating accuracy of vegetation NPP decreases in rugged terrain， due to the optical radiative transfer distortion caused by the topographic effect. In the report， in order to eliminate the effects of topographic shadow on vegetation NPP estimation， the shadow-eliminated vegetation index （SEVI） was used to improve the photosynthetically active radiation （FPAR） model of the Carnegie-Ames-Stanford approach （CASA） model and analyze the spatial-temporal distribution characteristics of vegetation NPP in Fujian Province of 2005 and 2015. The results showed that when the relative error of FPAR in shadow calculated by the SEVI information was reduced to 0.53%， the effects of topographic shadow were eliminated. With the improved method， the average vegetation NPP of Fujian province was 861.9 g·m^-2·a^-1 in 2005 and 855.7 g·m^-2·a^-1 in 2015， respectively. The NPP of evergreen broad-leaved forest was the highest， and that of farmland was the lowest. The distribution of vegetation NPP was high in the western inland area and low in the eastern coastal area. The overall trend of monthly average vegetation NPP was approximately to that of temperature， with the highest NPP in summer of above 110 g·m^-2， while the lowest NPP in winter of less than 20 g·m^-2.