The principal component analysis is a multivariate statistical procedure used to reduce a set of numerous variables into a few ones that can explain a great percentage of the original variability. In this article, the principal component analysis combined with geostatistical tools were used to characterize different zones from the watershed known as “Hoya del rio Suarez”, for differential agricultural management. Five physical properties were used from the watershed database: sand content in percentage, mean weighted diameter of soil aggregates, available water holding capacity, bulk density and particle density. There were used 932 points of observations from a grid of 70x70 m on the terrain. The two first principal components were used in the analysis of spatial variability. The first component was associated with the soil compaction, and it explained 40.5% of the total variance of the physical variable set. The second component was associated with soil erodibility, and it explained 21.7% of the total variance. The maps built using these components were useful to characterize the watershed soils for compaction and erodibility problems. This result will be useful to propose differential soil conservation management practices.