Forest plays an important role in regional and global carbon cycle. Accurate estimation of biomass and carbon densities of various forest types is important to assess their contributions to total carbon storage in a region. In this study, based on the national forest inventory data in 2005 and 2010, the biomass of main tree species of the forests in the southern Lüliang Mountains was estimated by using both the weighted biomass regression model (WBRM) and the continuous function for biomass expansion factor (CFBEF). The results showed that the estimates from the WBRM were significantly higher than those from the CFBEF (P<0.01), and the WBRM was better for biomass estimation at the medium to small scales compared with the CFBEF. On the basis of 28 × 112 (species × plots) carbon density matrix obtained by WBRM, the classification and ordination of forest communities were carried out using the methods of TWINSPAN, DCA and CCA, respectively. Meanwhile, oneway ANOVA was used to test the significance of difference in carbon density among different forest formations, and Pearson correlation analysis was used to assess the correlation of carbon density with the environmental factors (elevation, slope, aspect and position). The results showed that the forest communities in the southern Lüliang Mountains were classified into 8 forest formations, and a significant difference in carbon density was found among these formations (P<0.01). Form. Quercus wutaishanica + Acer mono and Form. Q.utaishanica + Pinus tabuliformis had significantly higher carbon densities than others, and the carbon density of Form. Pinus bungeana + Platycladus orientalis was the lowest among the eight formations. The carbon density in 2010 was significantly higher than that in 2005. The total carbon density increased with an average value of 1.54 t·hm^-2·a^-1. The impacts of the environmental factors on the formations were in order of elevation > slope > aspect > position. There were significant correlations between the carbon density and elevation, and slope. The carbon density was first increased with the increase of elevation/slope, and then decreased. The forest carbon density was higher on shady and halfshady slopes (north and east) than on sunny and halfsunny slopes (south and eastsouth), and was lowest on steep slopes. In addition to tending management, tree species with stronger adaptability to environmental conditions should be selected for reforestation in order to enhance forest carbon density.