To evaluate the genetic diversity and structure of the South Korean soybean population, 233 landraces collected in various regions of the country were surveyed for 15 allozyme loci and one protein locus. The South Korean population was fixed or nearly fixed at seven of the 16 loci tested. The number of alleles per locus was 2.06 and Nei’s gene diversity was 0.194. These values were lower than the values for the same 16 loci previously reported for the Japanese and Chinese populations. The differences among eight regional groups were not so marked, with only 7.2% of the total variation arising from regional differentiation. Three southern regional groups (Chollabuk-do, Chollanam-do and Kyong-sangnam-do) exhibited a relatively high variability because of frequent occurrence of alleles characteristic of the Japanese population. A marked difference was found in allelic frequencies at the Dial locus between large-seeded landraces and small-seeded ones, suggesting that the latter, which are used mainly for bean sprouts, had been established independently of the former, which are used mostly for soy sauce and cooking with rice. Not only the region but also the usage as food materials should therefore be taken into consideration in designing an efficient collection and preservation method for the Korean soybean landraces.