Abstract: Prediction of breeding values(BV) is the major component of the evaluation of forest genetics. In this paper, we introduced the main methods and principles predicting the breeding values in forest trees and with special reference to Larix kaempferi, and discussed their applications in the prediction of breeding values and the selection of optimal genotypes. The results indicate that breeding values could be precisely predicted by best linear unbiased prediction(BLUP) or best linear prediction(BLP) based on balanced or quasi-balanced data as well as unrelated and unbalanced data. BLP was a simpler method, while BLUP was most suitable when predicting breeding values for the materials that are related or have different genetic effects. Selection index would be an optimal approach for multi-trait selection. Finally, we discussed the applications and issues of these two methods in the forest genetic evaluation.