Abstract: Shrub biomass is an important part of forest biomass. Quantification of shrub biomass is important to understand the fixation, depletion, distribution, accumulation and translation of materials and energy in the whole forest ecosystem. Total harvesting is generally impractical or inappropriate in forest studies. So allometric methods have been developed to estimate biomass form nondestructive surrogate measurements such as canopy diameter and plant height. The objective of this paper is to find a simple, convenient and accurate method to estimate Loropetalum chinense biomass since L. chinense is one of common shrub species in subtropical forest in China. Statistics methods were used to choose the best models with canopy diameter (C) and plant height (H) as variables. The results showed the relationship coefficient between biomass and CH (canopy diameter multiply plant height) variable is the biggest and W=ha (CH) b1 is the best model for estimating biomass. The best estimate models for branch biomass, leaves biomass and upper biomass are W1=0.000796 (CH) 1.1878, W2=0.0114 (CH) 0.7581, W3=0.0024 (CH) 1.0973 respectively. Compared with the models with D2H (basal diameter square multiply plant height), the power curve models with CH as variable have the same estimation precision. With the convenience in real forest investigation, the models with CH as variable are better than the ones with D2H as variable. The high precision of the model with CH as variable indicated that L. chinense has columniform morphology approximately. The results should be validated whether they fit trees from a particular stand.