Main Article Content
This study was carried out in an estational semideciduous montana forest, located in the Forest Reserve at Federal University of Lavras. The main propose was to improve the prescription for forest management, comparing the linear and non-linear adjusting of diametric distribution. In order to generate the prescriptions, 17 species with greater population density were selected and the nonlinear and linear functions were fitted. After the frequency estimates by diameter class, the De Liocourt quotient q was used for obtaining the management prescriptions. Based on the results, it was observed that the estimated frequency, by nonlinear function and by specie, had values closer to observed frequency, generating lower residual standard error. Therefore, the nonlinear method is more precise and exact in estimating the wood stock of the species, generating better prediction of the production.