A model is shown which has no dominance or dominance types of epistasis but which can result in heterosis of a hybrid over the best parent. The heterosis is due, in this case, entirely to additive X additive epistasis. The fact that dominance affects inbreeding depress i on but that additive X additive epistasis does not s ugges ts that the amount of inbreeding depression is a better criterion for deciding whether or not to breed for hybrids in a particular crop.