There has been an increasing use of soil sampling grids for basing the recommendation of lime and fertilizers in a precision agriculture system in the west of Parana , despising the undulating topography . Thus, this study aimed to evaluate the variability of soil fertility and the soybean Trabalho selecionado durante a VI Semana de Ciencias Agrarias (VI SECIAGRA), realizada de 01 a 03/10/2012 yield on a farm in the west of Parana which uses no-till system , previously divided into management units in a precision farming system . The experiment was conducted in eutroferric Oxisol characterized by being extremely loamy, in a property of 17.80 ha in Guaira (Parana, Brazil) cropped with soybeans in the 2010/11 season . The mapping of the propert y was done by collecting geographical coordinates at key points through GPS , and soil samples were collected in eight management units defined by agronomic knowledge of the property. Subsequently, the variables pH CaCl 2 , Al +3 , H+ Al, Ca +2 , Mg +2 , K + , P, P-rem , C, Cu, Zn, Fe, Mn, SB, effective CTC , CTC, V% , Ca/Mg , Ca/K , Mg/K , ( Ca +Mg ) 2 /K and K/raiz 2 ( Ca +Mg ) were used to establish the dissimilarity matrix and build the map of the property arranging the most similar management units together. The productivity of the eight management units were also evaluated by using four replications in a completely randomized design , totaling 32 sampling units . The definition of management units by using agronomic knowledge and the regional edaphoclimatic conditions enable to capture the variability of soil fertility. There was the formation of four groups of management based on dissimilarity : group I ( management units 1 and 2) , group II (3 and 4 ), group III (5 and 8) and group IV (6 and 7). N o difference in soybean yields was observed in the eight management units.