The use of productivity level management zones (MZ) has demonstrated good potential for the site-specific management of crop inputs in traditional row crops. The objectives of this research were to analyze the process of defining MZs and develop methods to evaluate the quality of MZ maps. Two approaches were used to select the layers to be used in the MZ definition: 1) Statistical Approach (SA_MZ) and 2) Agronomic Approach (AA_MZ). The difference is that in the AA_MZ approach all non stable variables were discarded. Two field datasets were used in this study: 1) Field 1: a 17.0 ha soybean area located in Cascavel, Paraná, Brazil: and 2) Field 2: a 35.0 ha corn area located in Wiggins, Colorado, US. Two and three MZ were created using the k-means clustering technique. The proposed methodology for definition of management zones in this study was proved simple and allowed comparison of quality of generated MZs. Independent of the approach used to define management zones, each approach found that MZs could be used as yield goal map for preplant or in-season fertilizer recommendation for the precision crop management.