Japan is a historical center of genetic diversity for soybean [Glycine max (L.) Merr.], but diversity of modern Japanese cultivars is not well characterized. The objectives of this study were to quantify genetic diversity of Japanese cultivars via coefficient of parentage (CP), determine the relative importance of breeding factors in explaining that diversity, and incorporate results into a practical guide for management of diversity. All 86 public Japanese cultivars released and registered during 1950 to 1988 were subjected to CP and multivariate analysis. The mean CP for the 86 cultivars was low (0.04), indicating a potentially high degree of diversity in Japanese breeding. Eighty percent of all pairs of cultivars were completely unrelated by pedigree. The low mean CP for the cultivars was attributed to a continual incorporation of unique Japanese land races into the genetic base over time, to the introduction of foreign germplasm from China and the United States and Canada (US-CAN) as breeding stock, and to limited exchange of germplasm among Japanese breeding programs. Cluster analysis was an effective discriminator of diversity. Six clusters were identified which had a mean CP value equivalent to that of half-sibs or greater. These clusters encompassed a total of 54 cultivars, explained 57% of the variation in the CP relations, and had few ancestors in common. Each cluster was derived primarily from only a few programs. Backcrossing and full-sib matings were absent in Japanese pedigrees and, thus, clusters were formed primarily from parent-offspring, full-sib, and half-sib relations. Cultivar attributes such as growing region, release era, maturity designation, and developing institution did not elucidate strong patterns of pedigree diversity. In practical breeding, one may maximize the chances of finding good specific Japanese × Japanese or Japanese × US-CAN crosses by choosing Japanese cultivars from a wide array of Japanese clusters rather than sampling extensively within a cluster.