The mechanized sugarcane harvesting system without burning the trash changes several production environment characteristics, either because of intense machinery traffic or because the trash layer is maintained on the soil surface. The goal of this work was to evaluate harvesting system impacts on some soil physical attributes. The field research was conducted on a soil classified as a Typic Hapludox. Green cane mechanized harvest was compared against manual harvest after burning the trash. The experimental design was organized as swath treatments (harvesting systems), with 5 sampling points (replicates within harvesting systems). Each swath was 21 meters wide by 180 meters long, and spacing between sugarcane rows was 1.4 m. The surveys were carried out after three harvest cycles. Soil moisture and temperature were evaluated every 14 days, in the period between October/1999 and February/2000. The gravimetric method was used to determine moisture, on a weight basis. Samples were taken at depths of 0-20, 20-40, 40-60, and 60-80 centimeters and results were separated into data for dry periods and wet periods. We considered as dry period samplings those performed after 15 days, on average, from the last occurrence of precipitations higher than 5 mm, while wet period samplings were those performed until 2 days, on average, after precipitations higher than 40 mm (5 sampling dates). Temperature was measured using a 24X data logger, with thermocouple wires at depths of 3, 6, 12, and 24 centimeters. Density evaluation was achieved by collecting undeformed samples at 3 positions (under the sugarcane row, at 35 centimeters from the row, and at the interrows) and at 4 depths (0-20, 20-40, 40-60, and 60-80 centimeters). During the dry periods, the soil conserved more moisture in the area submitted to mechanized green cane harvest in relation to the area harvested by burning the cane trash. On average, moisture was 1.5% higher in the mechanized area, equivalent to 18 l m-3 more water (Figure 1). The wet periods maintained soil moisture higher than 13% (0 to 20 centimeters), in the average of the five sampling dates. On those dates, there was no significant difference in water contents between both harvesting systems, because water availability was high, near field capacity. Therefore, it is during critical drought periods that the residual trash layer from green cane harvest contributes to maintain soil moisture, thus attenuating the effects of dry and hot weather spells on plant metabolism. The mean daily soil temperatures under mechanical green cane harvest were smaller than under burned cane. The greatest differences were verified at the 6-centimeter depth in the months of November and December (up to 3.5 oC), and tended to decrease and become stabilized from the end of January/00 due to an increase in leaf cover (Figure 2). Smaller daily temperature amplitudes also occurred in the swath where cane was not burned, which is favorable for plant development. Because mechanized harvest operations generate more intense traffic than manual harvest, they resulted in higher bulk density until the 40 cm depth (Figure 3). The soil under mechanized green cane harvest showed higher density, even under the sugarcane-growing row. This effect, accumulated since the treatments were first implemented, is caused by the transmission of forces along soil particles. Other studies showed the effects of traffic on annual crops and verified compaction to the sides of wheeling tracks, in the soil portion under planting rows. The higher densities among the three positions were found at 35 centimeters from the cane-growing row, higher than at the interrow position, where subsoiling is performed during fertilization. The trash layer that is left after mechanized green cane harvest provides higher moisture, lower soil temperature, and narrower temperature amplitude when compared with the burned-cane harvesting system. On the other hand, high machinery traffic results in increased bulk density until the 40-centimeter depth. Project funded by FAPESP and developed at UNESP/Jaboticabal, Brazil.