One of the greatest challenges in plant breeding is the selection of parental combinations that will produce progeny that are superior to either parent. The objective of this research was to determine if yield heterosis in the F2 bulk of a soybean [Glycine max (L.) Merr.] single cross is a predictor of significant genetic variance for yield and superior line yields among random inbred lines derived from the F2 generation. In one experiment, thirty‐seven F6:7 inbred lines from each of five bi‐parental populations along with parents and the bulk F2 generation were yield tested at three locations. Of the five populations, three had significant genotypic variance for yield. The only two transgressive segregating lines came from the population with the greatest genotypic variation for yield and the only one with mid‐parent F2 yield heterosis (15% at P > 0.06). In another experiment, random inbred lines were derived from four single crosses chosen from a seven parent half diallel cross with similar mid‐parent yields. Two exhibited F2 yield heterosis and two did not. Derived F5:6 lines were yield tested in three environments. After adjustment for maturity, both populations derived from the non‐heterotic crosses had zero genetic variation. Genetic variance estimates for the two populations derived from heterotic crosses were larger than their standard errors and significant at P > Z = 0.07 and P > Z = 0.05. These experiments provide evidence that F2 bulk heterosis can be a predictor of significant genetic variation among inbred soybean lines derived from a single cross.