This paper focuses on the selection of height-diameter curve (HDC) which characterizes the relationship between tree height and diameter at breast height (DBH) for growth prediction. Tree height and DBH are commonly used variables in monitoring forest growth and predicting its stock. To select the appropriate HDC among multiple candidates, empirical rules or mathematical approaches based on the residual sum of squares have been applied in previous research. In this paper we apply cross-validation (CV) criterion to select an appropriate HDC for the purpose of forecasting. The CV criterion is easy to use because it is based on simple iterative process without any assumptions required on the candidate models for HDC. Not only is CV easy to use, it also evaluates forecast accuracy, which is consistent with the objectives for HDC use. In this paper, we demonstrate the results for analyzing real life data of sugi (Cryptomeria japonica) stands in Japan by preparing five candidates of HDC. We also show the results of numerical experiments for verifying the ability of the method introduced in this paper.