Biomass in different parts of individual trees was determined using the measured data from different life forms of trees in the shrub layer in Tiantong region. Biomass regression models developed for main shrub species were used to predict above-ground biomass. The accuracy of the models was estimated by comparing to the direct harvest method. The results showed that deciduous species distributed more biomass to above-ground part than evergreen species to receive more light resources to meet their growth needs. Shrubs distributed more biomass to the leaf, while trees accumulated more biomass in the stem and root. Biomass regression models for individual organs of main shrub layer species were almost power functions. Biomass models for tree trunk and above-ground parts fitted best. Thus, the model for tree trunk had the highest accuracy while the model for leaf biomass had the lowest accuracy. Therefore, our locally developed regression models can accurately estimate above-ground biomass of the shrub layer in the studied region only with an error of 0.47 % compared to the direct harvest data, indicating that the model is suitable to biomass estimation for the shrub layer in this region.