Abstract: As a key parameter in monitoring vegetation productivity by remote sensing driven model, rapid and accurate acquisition of vegetation light use efficiency (LUE) in large area has been a key problem. Selecting typical Phragmites australis wetland in Northeast China as study area, multitemporal Landsat OLI (operational land imager) image and the object-oriented classification method were used to extract Phragmites australis wetland. Based on the principle of vegetation physiology and ecology, the relationship among LUE, vegetation indexes and chlorophyll content was analyzed, the feasibility of accurate estimation of LUE for wetland vegetation by spectral vegetation index was discussed. Results show that areas of Phragmites australis wetland in Qixinghe Wetland, Chagan Lake Wetland and Shuangtai Estuary Wetland were 122.19, 75.29 and 439.61 km², respectively, and overall classification accuracy was more than 82%. With the exception of EVI (enhanced vegetation index), other six vegetation indices showed the same spatial pattern characteristics with those of three wetlands. Totally, values of vegetation indices for different land covers were: cultivated land > Phragmites australis > other wetland vegetation > water body. There exists close relationship among LUE, chlorophyll and vegetation index. NDVI (normalized difference vegetation index) was most sensitive to LUE (P < 0.01;R²=0.62), which was the best one to characterize LUE of Phragmites australis in this study. This study verified the theoretical hypothesis that LUE could be inversed efficiently by remote sensing vegetation index taking chlorophyll as the intermediate variable, which can provide references for the study of vegetation productivity and carbon cycle on regional scale.