Resumo. This research evaluated the influence of water content on soil compaction of an Entisol under traffic of forest harvesting machines and developed models to estimate soil compaction. The study was conducted in areas of a forest company, located in the state of Santa Catarina, Brazil. The sampling was performed under various conditions of soil moisture, contemplating three classes: Ug1 (Ug < 0,30 kg kg-1); Ug2 (0,30 ≤ Ug < 0,40 kg kg-1); Ug3 (Ug ≥ 0,40 kg kg-1). The soil bulk density was determined (Ds), total (Pt) and air-filled porosity (PA) and penetration resistance (RP), under the conditions without traffic (WT) and after traffic (AT) of a Feller-buncher (CAT 522) and a Skidder (CAT 545) employed in wood harvesting of Pinus taeda stands. Multiple regression equations were adjusted to estimate the soil penetration resistance, through arithmetic and logarithmic models with pure and mixed variables. The results indicate that in moisture condition above than 0,40 kg kg-1 (Ug3), the traffic promoted increase in Bd and PR of 10 and 24%, respectively, and decrease in αt and αair, in order of 12 and 43%. In low moisture content (Ug < 0,30 kg kg-1), the increase of Bd and PR was 3 and 13%, respectively, and the decrease of αair of 8% and without effect in αt. The stepwise logarithmic model with mixed variables (SLMM) showed the best results for the estimation of soil penetration resistance, but caution must be raised in their employment due to the possible influence of other factors in the estimation and determination of PR.