Heterosis and inbreeding depression are known to be opposite phenomena that depend on allele frequencies and directional dominance. Heterosis refers to the superiority of the hybrid over its parents by an increase in the mean of crossbred individuals, while inbreeding depression refers to the reduction in the phenotypic mean of a population. Heterosis has been described as a function of the squared of the genetic divergence in allele frequency of the parents (Δ2), and dominance (d), while inbreeding depression depends on d and allele frequencies. We derived a model of heterosis based on genetic divergence in allele frequencies between the parents (Δ), dominance, and inbreeding depression. Similarly, to better understand Δ and inbreeding depression we estimate shared identical by descent (IBD) segments between inbred lines of maize. The main objective was to understand the underlying basis of heterosis and to estimate genetic diversity and progenitor’s genetic contribution based on the amount of shared IBD segments. To describe heterosis, six synthetic maize populations and eight inbred lines were used. Three crosses between synthetic populations, three between synthetic populations and the B129 inbred line, and six between inbred lines were evaluated in nine environments under a modified split-plot design with three replications. For easy deductions, heterosis model was defined under a single-locus two-alleles case and tested using a “goodness-of-fit” test. To estimate IBD segments, a set of 44 ex-PVP lines along with eight key ancestors of maize in the U.S. Corn Belt were selected. Shared IBD segments were identified by using a probabilistic approach based on a Hidden Markov Model (HMM) framework. Genetic diversity between individuals was estimated as 1 minus the kinship coefficient. Genome-wide kinship coefficients were calculated from the posterior probability of the IBD status at each locus. Our results showed that a single-locus two allele model of heterosis was adequate to describe the variation in the mean of each generation and to predict heterosis. Heterosis estimates were significantly higher for crosses involving inbred parent populations, with a limiting value when the parents reach complete inbreeding. Both model and empirical evidence of heterosis shows that population divergence in allele frequency between parents (Δ) is the key driver of heterosis, but that this divergence is achieved when the level of heterozygosis within each population decreases (i.e. increased FST). Therefore, there was a negative relationship between midparent heterosis and inbreeding depression, the latter expected to be high when panmictic populations are used and low when there is no more heterozygosity in the parents. Hence, having a deeper understanding of inbreeding could lead to better predictions of heterosis. For the 44 ex-PVP lines, long IBD segments (>14.5 Mb) were predominant between ex-PVP and key ancestor lines, suggesting a recent inbreeding, originated in less than 15 generations. There was a high genetic diversity between heterotic groups (stiff stalk and non-stiff stalk), with a reduced diversity among lines in the stiff stalk group. Consequently, we found that a small group of ancestors have contributed large proportions of the genome to important PVP lines in the U.S. Corn Belt. Finally, our results provide a high-resolution data analysis that helps in the identification of IBD regions in the genome that could be used in the quantification of the degree of divergence between parents (Δ), constituting a way to identify the best combination of parents to maximize heterosis.