A total of 2959 Chinese cultivated soybeans were examined for saponin composition by thin-layer chromatography. Thirteen components were distinguished as five groups, group A (Aa, Ab, Ax, Ad, Ae and Af), DDMP (αg and βg), group B (Ba), group α (H-αg, I-αg and K-αg), and group β (KA-αg). The detected composition included four common phenotypes (Aa, Ab, AaBc, AbBc) and five new types (AaβKA, AbαH, AaαI, AbαI, and AbαK). The group α components and group β were first found in cultivated soybeans. The frequencies of the Sg-1 a and Sg-1 b alleles for Xyl and Glc at the C-22 terminal sugar were 39.3 and 60.7%, respectively, and the Sg-4 gene for Ara at the second sugar of C-3 was rare (0.07%). The frequency distribution in botanical characters indicated that the Sg-1 locus was related to soybean evolution and ecotypes; small seed, prostrate stem, black seed-coat, indefinite growth, broad-leaved shape, low-oil content, and spring-sowing types accumulated relatively higher frequencies of Sg-1 a allele. The frequency of Sg-1 a allele was regionally higher (>60%) along the Yellow River valley. Integrating archaeological documents, the frequency distribution of Sg-1 a allele and the occurrence of group α in soybeans with a relatively higher frequency in the northwest, a single area of origin for soybeans is proposed as the middle and lower reaches of the Wei River valley, i.e., the middle-upper reaches of the Yellow River valley in China. Group α components in soybeans are safe for humans because these saponins have been ingested for several 1000 years in China.