Abstract: 【Objective】Accurately estimating the biomass of seedlings, saplings and shrubs is essential for forest ecosystem management.【Method】A total of 39 woody species with 746 individuals in the Jiulian Mountain Nature Reserve located in southeast China were destructively sampled. All species were categorized as high trees, small trees and shrubs, based on their life form. In addition to component-specific biomass, the ground diameter(d), stem height(h)and basic wood density(ρ)were measured, and regressive models were established with d², ρd², d²h and ρd²h as predicting variables, respectively. Optimum models were chosen based on the determination coefficient R² and standard error of estimate.【Result】The basic wood densities of the sampled species were in the range of 0.459-0.784 g/cm³. Sixteen optimum models were chosen among 64 candidate models. For small trees and shrubs, equations for foliage and branches with a single predicting variable d, showed higher R² values, while the fitness of the stemand total-aboveground biomass was closer for equations with d, h, and ρ as the independent factors for small trees and shrubs than for the other category.【Conclusion】The models established in this study are expected to provide reliable accuracy for estimating the above ground biomass of understory woody species in the same vegetation zone.