Knowledge of the spatial variability of soil fertility is very important, especially in precision agriculture. Thus during the 2009/2010 agricultural year, in the city of Selva (MS) in the Brazilian Cerrado, the objective was to analyse the spatial variability of the fertility attributes of a Dystropheric Red Latosol under a no-tillage system. A sampling grid was installed to collect soil samples, with 120 sampling points in an area of 3.0 ha and on an even slope of 0.055 m m-1. The available phosphorus content, organic matter content, values of pH (CaCl2), levels of K+, Ca+2, Mg+2, values of base and aluminum saturation were all determined at depths of 0 to 0.10m and 0.10 to 0.20m. A classic descriptive analysis was carried out with the help of the SAS statistical software, and semivariograms were then modeled for all attributes, giving the respective cross-validations and kriging chartswith the GS+ 7.0 software. The largest variability, in the chemical attributes analysed by the coefficient of variation, occurred in the soil-depth layer of 0.10-0.20m. All studied chemical attributes showed spatial dependence, it being possible to map the study area. As they represent the spatial continuity of the semivariograms obtained in this study, the values of the geostatistical ranges recommended for those attributes here studied, should be between 40.2 and 113.1 metres.
