Weathering deterioration in the pre-harvest phase impairs high-quality soybean seed production. The performance of several tests to infer seed quality is common in genotype selection, and multivariate statistics can assist in the interpretations. This study aimed to assess the efficiency of the principal component analysis (PCA) and canonical discriminant analysis (CDA) multivariate statistical methods in assessing the tolerance of seeds of different soybean cultivars to weathering deterioration in the pre-harvest phase under greenhouse conditions. Seeds of six soybean cultivars (DM 6563, BMX Apolo, BMX Potência, NA 5909, NS 5959, and TMG 1175) were produced. Different simulated precipitation levels (0, 60, 120, and 180 mm) were applied in the pre-harvest phase. The seeds were collected and assessed for physiological, physical, and biochemical analyses and the data were analyzed by PCA and CDA techniques. The results showed that PCA and CDA are efficient for assessing the tolerance to weathering deterioration in soybean seeds. PCA and CDA assisted in the recommendation of the tests first germination count, accelerated aging, tetrazolium, percentage of seeds with seed coat wrinkling, protein content, and protease activity in the pre-selection of genotypes for weathering deterioration. PCA and CDA also helped to identify the cultivars DM 6563 and BMX Potência as more susceptible and NA 5909 and TMG 1175 as more tolerant to weathering deterioration in the pre-harvest phase.