【Objective】The prediction model of bark factor and bark thickness at any height of larch plantation was established in order to predict bark thickness more accurately and provide more accurate prediction model and guidance for actual wood production and forest management.【Method】Based on 1 186 disk data of 49 artificial larch trees in Mengjiagang Forest Farm of Jiamusi City, Heilongjiang Province in 2015, the bark thickness (bark factor, bark thickness at any height) of larch plantation linear mixed effects prediction model was constructed using the MIXED module in SAS 9.4 software. The model evaluation indicators were Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), -2 log likelihood (-2LL) and likelihood ratio test (LRT). 【Result】For the bark factor model, the bark factor model with b₁,b₂, b₄ and random parameter combination is the optimal mixed model based on the tree effect, and the model with b₁, b₂ random parameter combination is the optimal model based on the plot effect. For the bark thickness model at any height, the combination of b₁, b₂ is the optimal mixed model based on the tree effect, and the combination of b₀,b₂,b₃ is the optimal model based on the plot effect. All the optimal models have the best fitting effect when they have unstructured (UN) variance-covariance matrix. 【Conclusion】The tree effect has the greatest influence on the model whether it is the bark factor or the bark thickness model. The prediction accuracy of the mixed-effect model is significantly improved compared with the traditional regression model.