The characterization of the physiological potential of soybean seed lots is accomplished by germination and vigor tests, which estimate the capacity of the seeds to establish an adequate field population. However, the choice of a single test does not reflect the capacity of the seeds to establish an adequate field population, because every test assesses the seeds in a different way, sometimes under optimal environment conditions, sometimes under stressful conditions. Multivariate analysis is a technique that permits to understand the dependence structure contained in the variables and to characterize groups of samples under a specific standards. The aim of this work was to discriminate lots of soybean seeds, using variables obtained from germination and vigor tests. It was used 10 lots of each soybean seeds (cv. BRS Valiosa and cv. Msoy 7908).. Initially, it was evaluated the physiological potential of seeds using the germination and vigor tests (accelerated aging and electrical conductivity) as well as seedling field emergence. Also the germination was determined under different soil water content, based on the percentage of the total soil porosity occupied by water. All seed lots were processed by cluster analysis by hierarchical method as a measure of similarity between pairs of seed lots, considering Euclidean distance and UPGMA method (Unweighted Pair Group Method with Arithmetic means) as the connection method between the groups. Then, the data were submitted to principal components analysis to detect those variables with higher capacity for discrimination. Cluster analysis allowed the stratification of the lots in groups, while the principal component analysis showed that the field emergence, accelerated aging and germination under different soil water content are more associated with cultivar BRS Valiosa and the germination and electrical conductivity are more associated with Msoy 7908.