The objective of this study was to understand the effect of the spatial variability of the soil physical attributes in areas cultivated with cassava with different soil tillage systems, using the techniques of classical statistics and geostatistics. The experiment was carried in 2013 in São Mateus - ES, Brazil, on a cohesive ultisol, with a plan relief. The preparation of the soil for transplanting were made in two adjacent areas of approximately 0.5 ha each, using up to two passes of a harrow disk or two passes of a chisel plow. The following soil physical attributes were studied 90 days after transplanting: soil density (DS), macroporosity (Macro) and microporosity (Micro), soil water content according to two preparations systems and two depth ranges. Eighty plants were randomly selected to perform the experiment. The soil physical properties resulted in spatial variability due to the strong dependence for all variables, tillage system and soil depth. The semivariograms were adjusted to the spherical and exponential models for the evaluated physical attributes. Except for water content, it was observed an increase in the range (A0) with increasing depth for both tillage systems. The soil water content presented a decrease about 23% for the harrowing and scarification systems.The objective of this study was to understand the effect of the spatial variability of the soil physical attributes in areas cultivated with cassava with different soil tillage systems, using the techniques of classical statistics and geostatistics. The experiment was carried in 2013 in São Mateus - ES, Brazil, on a cohesive ultisol, with a plan relief. The preparation of the soil for transplanting were made in two adjacent areas of approximately 0.5 ha each, using up to two passes of a harrow disk or two passes of a chisel plow. The following soil physical attributes were studied 90 days after transplanting: soil density (DS), macroporosity (Macro) and microporosity (Micro), soil water content according to two preparations systems and two depth ranges. Eighty plants were randomly selected to perform the experiment. The soil physical properties resulted in spatial variability due to the strong dependence for all variables, tillage system and soil depth. The semivariograms were adjusted to the spherical and exponential models for the evaluated physical attributes. Except for water content, it was observed an increase in the range (A0) with increasing depth for both tillage systems. The soil water content presented a decrease about 23% for the harrowing and scarification systems.