Resumo. The objective of this study was to determine the most suitable number of discs, the best sampling positions along the stem and to propose equations to estimate the basic density of trees using a reduced number of discs in various phytophysiognomies, in order to subsidize biomass estimates. The detailed scaling data of the trees were obtained from the Forest Inventory of Minas Gerais state, and is composed by 1,325 trees distributed in regions, diameters and height classes and various forest formations. We took five discs from each at positions 0 (A), 25 (B), 50 (C) 75 (D) and 100% (E) of commercial height. The density of each disk was determined according to ABNT NBR 11.941 (2003) and the overall stem density from the arithmetic mean of the disc densities. Analyses were performed by regression analysis, the Akaike Information Criterion, Standard Error of the Estimate and graphical analysis of residuals. The results indicated that, in general, the best sampling position for estimating the basic density of stem wood with only one disc should be performed in the middle of the main stem, with two discs in positions 25 and 75% in height and with three discs at positions 0, 50 and 100% of the stem height. We also observed that the best estimate of stem density will always be given by the largest number of discs. However, it is possible to estimate the density of the trees with high precision, with a smaller number of discs provided that they are sampled at specific positions of the stem and inserted in the Equations according to each phytophysiology.