Abstract: In order to test the fractal network model and other three allometric models, the allometric relationship between stem diameter (D) and tree height (H) in Quercus variabilis plantations from the Sizuolou Nature Reserve of Beijing was studied based on the power function H=aDb. The results of reduced major axis (RMA) regression showed that in 12 out of 16 plots, the RMA slopes between logD and logH supported the predicted value (2/3) of elastic similarity and fractal network model, while only 2 plots supported the geometry similarity model and no data supported the stress similarity model; both univariate and multivariate analyses showed that habitat and stand factors did not have significant influence on DH allometry. This study suggests that the elastic similarity and fractal network model are more suitable for describing the allometric relationship between D and H than geometry similarity or stress similarity model.