The ideal agricultural soil, being a dispersed three-phase system, must have a volume of total porosity of 0.500 m3/m3, respectively with 0.355 and 0.165 m³/m³ of microporosity and macroporosity; as well as 0.500 m³/m³ at volume of solids. The bean grain yield (GY), under no-tillage system and irrigated, and the following soil attributes (total porosity-TP, macroporosity-MA, microporosity-MI, and the bulk density-BD), in a Dystroferric Red Latosol (Haplic Acrustox), located at the Experimental Station of the Faculdade de Agronomia/UNESP, in Ilha Solteira county, Sao Paulo State, Brazil (22º23' latitude S; 51º27' longitute W), were analyzed in the crop year 2004/2005. The purpose was to improve directresses related to the improvement of the crop productivity at issue. A geostatistical grid to collect soil and plant data was installed, with one hundred and nineteen sample points, in an area of 0.99 ha. All researched attributes showed low variability with the exception of the MA4, which it was medium. The linear correlation among the pairs of attributes showed: a) the increase of BD caused expressive decreases in the MA and TP, and b) the increases of MA and TP provide, although with low correlation, the decrease in the GY. Similary, looking at each attribute in private, the statistical analysis showed a spatial dependence varying between strong and moderate, with ranges between 22.9 and 77.9 m. However, when analyzed conjunctly, it was observed a satisfactory and harmonious inverse relationship between the TP and GY, occurring the same thing between the MA and GY. Thus, in the areas with the greatest values of the TP and MA occurred the smaller at the GY. So, the conclusion related to the linear and spatial correlations was that with the increasing of the macroporosity as well the total porosity the bean grains yield has decreased.