Based on the images of post atmospheric correction reflectance (PAC)，top of atmosphere reflectance (TOA)，satellite radiance (SR) and digital number （DN） of a SPOT5 HRG remote sensing image of Nanjing，China，two vegetation indices (VI)，i.e.，normalized difference vegetation index (NDVI)，and ratio vegetation index (RVI) were derived，and compared with the leaf area index (LAI) data acquired from field measurement. A total of 157 LAI-VI relationship models were established. The results show that LAI was positively correlated with VI (r=0.303~0.927，p<0.01). Independent variables of the optimal models corresponding to various vegetations included 2 vegetation indices at 3 radiometric correction levels，indicating potentials of vegetation indices based on different radiometric correction levels in LAI remote sensing retrieval. These optimal models included broad-leaf forest： LAI=-3.345+5.378RVI_SR+7.329NDVI_SR（R²=0.818，RMSE=0.527）；conifer-broad-leaf forest：LAI=1.696+17.076NDVI_DN+137.684（NDVI_DN）²-288.240（NDVI_DN）³（R²=0.919，RMSE=0.440）；shrub：LAI=-0.065+19.112NDVI_SR-113.820（NDVI_SR）²+184.207（NDVI_SR）³（R²=0.900， RMSE=0.448）；grass： LAI=-5.905+6.446RVI_SR+9.477NDVI_SR（R²=0.944， RMSE=0.378）；and total vegetation：LAI=-1.615+7.199NDVI_DN+2.640NDVI_SR+2.105RVI _PAC （R²=0.801， RMSE=0.668）. The study demonstrates that LAI remote sensing estimation of various types of vegetation based on images of different radiometric correction levels contributes to tapping of valuable information from remote sensing images，thus improving accuracy of LAI estimation.