By using retrieved LAI from SPOT remote sensing data, the relationship between leaf area index and the normalized difference vegetation index was studied. The regular pattern of LAI in deference growth stages was combined to estimate models between LAI from remote sensing data and yield of sugarcane. After optimizing the models, the best model for sugarcane yield estimation was determined. The results showed that a strong positive correlation between LAI and NDVI was obtained. A quadratic function model was the best regression model for the whole growth stage (R2=0.8429). Statistical yield were compared with yield simulated with LAI. The relative error was 2.6%. The estimation of sugarcane yield estimates could be improved by combining remotely sensed data. This study provided the reference for estimating the regional yield of sugarcane in China.