In order to find an efficient estimation method of potato biomass, the potato biomass and corresponding hyperspectral data of the seedling stage, the tuber formation stage, the tuber growth stage, the tuber growth later stage and the starch accumulation stage in Beijing Xiaotangshan area were acquired during 2017. The R² contour maps of the coefficient of determination between vegetation index NDVI and RVI and biomass were constructed. The gravity center formula was used to analyze the sensitive bands of different regions to biomass. Finally, empirical regression analysis was used to estimate the dry potato biomass with the sensitive bands as independent variables. The results indicated that NDVI（382, 669） and RVI（385, 668） constructed by sensitive bands estimated accuracy reached extremely significant correlation at the level of P＜0.01,the modeling R² was 0.545, the validation accuracy RMSE was 0.054 kg/m², and MAE error was 0.040 and 0.039 kg/m². The research showed that among all regression analysis models, the parabola model was the best（R²=0.545）. The center of gravity of the R² maximum value region fitted by vegetation index NDVI and RVI can be used as the center of the sensitive band to improve the stability and accuracy of the model, provided a new band selection method for rapid nondestructive diagnosis of potato aboveground dry biomass.