The variation in plant functional traits can help us understand adaption mechanisms to different environments and predict the distribution of species and environmental changes. Twenty-one, 20 m×20 m plots were established and divided into 336, 5 m×5 m subplots in the tropical cloud forests in Bawangling, Hainan Island. Functional traits (i.e., leaf area[LA], leaf dry weight[LDW], leaf mass per area[LMA], leaf chlorophyll content[Chl], leaf thickness[LTh], and wood density[WD]) of all trees and shrubs with diameter at breast height over 5 cm, as well as soil nutrient contents, were measured in all plots. The magnitude of variation in each functional trait at individual plant, within-species, among-species, and community levels were assessed using a generalized linear model, whereas the relationships between soil nutrients and functional traits were examined using a stepwise linear regression. Results showed that the ranges of explained variance for LDW, LMA, CHl, LTh, LA, and WD were 0.06-0.47, 0.09-0.35, 0.35-0.72, and 0-0.07 for the individual, within-species, among-species, and community levels, respectively. This indicates that the magnitude of variation in functional traits was the highest at the among-species level and the lowest at the community level. The stepwise regression analysis showed that the changes of functional traits at all studied scales were significantly correlated with soil organic matter, nitrogen, and phosphorus.