In current research of plant functional traits, classifying plants into different growth forms to statistically analyze features of plant functional traits is a common approach. Different classifications of growth form, however, may lead to statistical bias of plant trait values, and such bias has not been evaluated. In this paper we investigate how different classifications of growth form influence the calculation of plant functional traits. We classified the growth form of 35 dominant woody plants in eastern subtropical evergreen broadleaved forest region in Beishan Mountain of Jinhua, Zhejiang Province into three schemes according to the Flora of China and actual situation in filed investigation: the classification 1, traditional tree and shrub described in the floras; classification 2, trees, small trees and shrubs based on height and diameter in breast height; and classification 3, trees and shrubs of tree layer and shrub layer according to height only. The statistical bias of plant leaf and twig traits under three classifications of growth form was analyzed and compared. The results showed that: (1) compared with the traditional classification 1, classification 2 had a significant influence on twig and leaf traits of trees than classification 3, while the influence on twig and leaf traits of shrubs was lower than classification 3. However, there was no significant difference of tree and shrub traits in different growth classification schemes. Compared with small trees in classification 2, leaf and twig traits of trees and shrubs in classification 1 and even within the classification 2 were significantly different to those of small trees in classification 2; (2) After subdividing different growth forms into different life forms, the impact of different growth form classification schemes on trait statistics increased. However, leaf and twig traits of both evergreen and deciduous small trees were still significantly different from those of trees and shrubs in different life forms. In general, different classification schemes of plant growth forms can result in the statistical bias of plant functional traits. Separating the small tree from the tree and shrub functional groups could better reflect the differences of plant functional traits in forest ecosystems.