The endosperm has played a significant role in the evolution of angiosperms because of its physiological and genetic relationships to the embryo. One manifestation of this evolutionary role is its abnormal development in interploidy crosses. It is now established that the endosperm develops abnormally in interploidy-intraspecific crosses when the maternal: paternal genome ratio deviates from 2∶1 in the endosperm itself. We propose an Endosperm Balance Number (EBN) hypothesis to explain endosperm development in both interploidy-intraspecific and interspecific crosses. Each species is assigned an EBN on the basis of its crossing behavior to a standard species. It is the EBN which determines the effective ploidy in the endosperm of each species, and it is the EBNs which must be in a 2∶1, maternal:paternal ratio. The EBN of a species may be determined by a few genes rather than the whole genome. This hypothesis brings most intraspecific-interploidy and interspecific crossing data under a single concept with respect to endosperm function. The implications of this hypothesis to isolating mechanisms, 2n gametes, the evolution of disomic polyploids, and reciprocal differences in seed development are discussed.