The objective of this study was to identify site-specific management zones by means of chemical soil fertility indicators on sugarcane-cultivated Latosols (Oxisols), using indicator kriging, aiming an optimized use of precision agriculture techniques. The study was conducted in a 90 ha section of a larger field (1.900 ha) near Jaboticabal, Sao Paulo State - Brazil (21 degrees 15'S and 48 degrees 18'W). A 50 m regular grid was drawn on the experimental area (300 by 3.000 m). Soil samples were obtained at each 50 m grid point (420 points) from the top 0.2 m. Organic matter, P, K and V values of the soil samples were determined. The soil data were coded as 0 or 1 when higher or lower than the threshold value established for each variable, respectively. Using the results of geostatistical analysis for the coded data set and indicator kriging, counter maps of probability of meeting the critical threshold value for each of the individual soil attributes and their combinations were drawn. It was possible to identify areas of different soil fertility levels by this method.