Choice of donor to improve an elite single cross maize (Zea maize L.) hybrid is critical to the success of a breeding program. Theory has been developed to identify lines (I-w) possessing favorable dominant alleles not present in the elite single cross (I-1 x I-2). This paper evaluates the probability of I-w to improve I-1 x I-2. Assuming complete dominance, mu G' estimates the relative number of loci with favorable alleles in I-w but not in I-1 or I-2. The estimates mu D' and mu F' are the relative number of loci with favorable alleles in I-1 or I-2, respectively, but not in I-w. Thus, the probability of a net gain of favorable alleles (PNG(g)) from I-w = max[mu G'/(mu G' + mu D'),mu G'/(mu G' + mu F')], which estimates the relative number of loci where favorable alleles can be gained as a proportion of the relative number of loci where favorable alleles can be either gained or lost. The PNG(g) can be estimated from phenotypic values of I-1 x I-w, I-2 x I-w, I-1 x I-2, I-1, and I-2. I-w per se need not be evaluated. If partial dominance and/or complementary epistasis is prevalent, mu E' and mu C' estimate additional loci where favorable alleles could be gained from I-w to improve I-1 and I-2, respectively, while mu D', mu F', and mu B' estimate additional loci where favorable alleles could be lost from I-1, I-2, or both I-1 and I-2, respectively. Thus, the probability of a net gain of favorable alleles (PNG(ceg)) from I-w = max[(mu E + mu G)'/ (mu B + mu D + mu E + mu G)',(mu C + mu G)'/(mu B + mu C + mu F + mu G)']. In this case, I-w needs to be evaluated. The statistics PNG(g) and PNG(ceg) can be used to estimate the genetic probability of success of a donor (I-w) to improve an elite hybrid.