Irrigation is essential for reliable crop in arid and semiarid regions. However, current irrigation practices may severely damage the cropland causing negatives effects on yield production and on soil and water quality. Better understanding of the processes taking place under irrigated cropping systems can be useful to reduce the problems described above. For this reason, the statistical multivariate procedures were used to define the patterns of similarity of physical and chemical soil properties and to analyze their effects on crop yield in different semiarid soils. Soil profiles were collected at different time periods to perform statistical analysis. Results showed that the use of numerical soil taxonomy based on similarity parameters is useful to identify and elaborate maps of environmental soil zoning and to develop sustainability indicators in agricultural production. Indeed, this procedure can give support to studies of environmental impact being an useful tool for planning policies.