Resumo. Traditionally, most studies use linear models to adjust the hypsometric relationship using diameter at breast height (DBH) as the only independent variable. However, it is known that the linearity does not characterizes faithfully biological growth and other variables related to forest stand, or individual trees, influence in height growth. The aim of this study was to evaluate linear and nonlinear models’ performance for hypsometric relationship of two tropical pine species, as well as the effect of covariates adding in models’ parameters estimation. The data used provided from Pinus caribaea var. hondurensis and Pinus oocarpa stands, with different ages and site index, located in the counties of Nova Ponte and Indianápolis, Minas Gerais state. Linear and nonlinear models were fitted in its original form and including the covariates age, site index and species in the estimation of each models’ parameter. The models were compared using the following statistical criteria: residual standard error (Syx), Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC) and also by graphical analysis of standardized residuals. It was found that nonlinear models show greater ability to describe height behavior, and the inclusion of covariates in models’ parameters improved its fit quality, considering the statistical accuracy criteria adopted.