It is an important academic issue in field vegetation ecology that selecting the optimum size and number of sample as controlled by a standard error of mean. In this paper, based on time and labor consumed and edge effect considered, the selection of optimum size and number of the sampled quadrats in estimating forest undergrowth biomass was studied under the standard error of mean needed to control. An optimum independent variable was chosen to simulate prediction equation of estimating biomass of forest undergrowth vegetation. The results indicated that 0.25 m² was the optimum size for sampling the undergrowth vegetation biomass in this site by using Wiegert’s method. However, the 0.25 m² quadrat was too small and may induce a large edge effect to make an overestimate of undergrowth biomass. Furthermore, it is difficult to harvest the undergrowth vegetation by using the small simple size since lots of shrubs were too large to be encircled in it. Considering the impacts of edge effect and variation of the biomass relative mean value, we suggested the optimum size of 2 m×1 m for undergrowth biomass sampling and 10 quadrats with the same size may control the standard error within 10% of the biomass average value. Of the three selected combination variables (D²H, PH, and CH) for establishing biomass estimation models, CH is the most suitable for undergrowth shrub biomass estimation than the others. However, a linear regression equation combined with the independent variable PH can enhance the efficiency and convenience of investigation. The optimum estimation equation with the independent variable PH for undergrowth herbage layer biomass in unit area were W_U＝11.65+4.25（PH）for aboveground biomass and W_D=24.23+6.85（PH）for the below ground biomass.