Abstract: According to the 31 analyzed trees, Logistic and four-parameter Richards equations were used to fit the height growth models of the dominating stems, the average stems and the overtopped stems of natural Larix gmelinii forests. All models were highly significant at 0.01 level. Correlations among the age(A), the DBH(D), the tree height(H), the live height under branches(h₁), the dead height under branches(h₂)were analyzed and verified that there was a significantly positive correlation among D, H, h₁ and h₂ at 0.01 level. Live height under branches and dead height under branches significantly fit the model at 0.01 level. Results indicated that Logistic equation was better than the four-parameter Richards equation when fitting the growth, logistic equation was evidenced by their corresponding model determinant coefficients R² of 0.829, 0.758, 0.807 for the dominating stem, the average stem and the overtopped stem respectively. In contrast, 0.771, 0.668, 0.824 were respectively derived by the four-parameter Richards equation. Obviously, two models were jointly well-performed when fitting the growth of the dominant trees and the overtopped trees compared to the average trees. After test, the measured values and the modeled values in terms of the growth of the dominant trees had statistically no significant difference(p>0.05), which indicated that the established models could be practical.