Seed saponin composition of 3025 wild soybean (Glycine soja Sieb. and Zucc.) accessions collected from nine regions of Korea was analysed by thin‐layer chromatography to determine its polymorphic variation and geographical distribution and find mutants in saponin components. The saponin composition of seed hypocotyls was primarily divided into seven phenotypes, designated as Aa, Ab, AaBc, AbBc, Aa+α, AaBc+α and AbBc+α. The predominant phenotypes were AaBc (55%), Aa (33%), AaBc+α (7.5%) and Aa+α (3.3%). The frequencies of Ab, AbBc and AbBc+α were very low (0.3‐0.5%). Codominant alleles Sg‐1 a and Sg‐1 b and dominant allele Sg‐4 occupied 98.6, 1.1 and 63.3%, respectively. Alleles Sg‐3 and Sg‐5 were found to be dominant in all the analysed accessions except the mutants. Three accessions were discovered as mutants via LC‐PDA/MS/MS. The accession CWS0115 did not produce saponin Aa and Ax, CWS2133 did not produce saponin Aa and Ab and CWS5095 did not produce any group A saponins. These newly determined mutants might be utilized in producing a new soybean variety with good taste as well as in biosynthetic studies.