Biomass of natural Castanopsis fissa community was measured using the clear cutting method in a 750 m² plot at the Xiaokeng of Nanling Mountain, Southern China, where 43 tree species grew and the Castanopsis fissa is distributed as predominant species. We obtained the dried weight of stem, branch and leaf, and measured tree height (H) and diameter at breast height (DBH or D) for all 267 trees with DBH above 2.0 centimeter. Aboveground biomass (AGB) of this forest community and its distribution among different layers, tree species and organs of tree layer were analyzed, respectively. The regression models of biomass were established based on these AGB data. The results suggested that total aboveground biomass of natural Castanopsis fissa community is estimated to be 131.149 t/hm². Biomass of tree layer, understory layer, liana and litter layer were 129.895 t/hm², 1.563t/hm², 0.267 t/hm², 2.424 t/hm², respectively. Tree stem, branch and leaf accounted for 85.0%, 10.6% and 4.4% of total biomass in tree layer, respectively. Biomass of dominant tree species, Castanopsis fissa and C. carlesii occupy about 46.3% and 9.8% of total biomass of tree layer, respectively, which suggests that biomass of forest community in early succession phase is mainly allocated to a few tree species. The allocation of total aboveground biomass for each DBH class is about 1.0% for <5 cm, 13.1% for 5-10 cm, 52.2% for 10-15 cm, 26.4% for 15-20 cm, 4.6% for 20-25 cm, and 2.7% for >25 cm DBH classes, respectively. The distribution of biomass pattern among different DBH classes indicated that the forest community was in early succession phase. Using D as independent variable, total biomass regression model of single tree species (Castanopsis fissa) is expressed as W_tagb=0.086D^2.538 (R² = 0.947). The estimated biomass values by this biomass model is 0.3% lower than the measured values by clear-cutting method; The biomass model using D²H as independent variable is expressed as W_tagb=244.79(D²H)^1.037 (R² = 0.968). The biomass estimated by this model was about 1.7% lower than the measured biomass values. Total biomass regression model using D as independent variable for 43 tree species of natural Castanopsis fissa community is expressed as W_tagb=0.116D^2.384 (R² = 0.934). The biomass estimated by this model is about 5.0% lower than the measured biomass values. The model using D²H as independent variable is calculated as W_tagb=184.274(D²H)^0.881 (R²=0.952). The model seems to underestimate about 6.9% than the measured biomass values. Therefore, we suggested that the biomass regression model of Castanopsis fissa community using D as independent variable had a relatively high precision and more practicality.