Paleopolyploidy is the result of genome duplications which occurred at least several million years ago (MYA). Such an event could either double the genome of a single species (autopolyploidy) or combine those of two species (allopolyploidy). Because of functional redundancy, genes are rapidly silenced or lost from the duplicated genomes. Most paleopolyploids, through evolutionary time, have lost their polyploid status through a process called diploidization, and are currently considered diploids e.g. baker's yeast, "Arabidopsis thaliana", and perhaps humans.

Paleopolyploidy is extensively studied in plant lineages. It has been found that almost all flowering plants have undergone at least one round of genome duplication at some point during their evolutionary history. Ancient genome duplications are also found in the early ancestor of vertebrates (which includes the human lineage) and another near the origin of the bony fishes. Evidence suggests that baker's yeast ("Saccharomyces cerevisiae"), which has a compact genome, experienced polyploidization during its evolutionary history.

The term "mesopolyploid" is sometimes used for species that have undergone whole genome multiplication events (whole genome duplication, whole genome triplification, etc.) in more recent history, such as within the last 17 million years.

The hypothesis of vertebrate paleopolyploidy originated as early as the 1970s, proposed by the biologist Susumu Ohno. He reasoned that the vertebrate genome could not achieve its complexity without large scale whole-genome duplications. The "two rounds of genome duplication" hypothesis (2R hypothesis) came about, and gained in popularity, especially among developmental biologists.

Some researchers have questioned the 2R hypothesis because it predicts that vertebrate genomes should have a 4:1 gene ratio compared with invertebrate genomes, and this is not supported by findings from the 48 vertebrate genome projects available in mid-2011. For example, the human genome consists of ~21,000 protein coding genes according to June, 2011 counts at UCSC and Ensembl genome analysis centers while an average invertebrate genome size is about 15,000 genes. The amphioxus genome sequence provided support for the hypothesis of two rounds of whole genome duplication, followed by loss of duplicate copies of most genes. Additional arguments against 2R were based on the lack of the (AB)(CD) tree topology amongst four members of a gene family in vertebrates. However, if the two genome duplications occurred close together, we would not expect to find this topology. A recent study generated the sea lamprey genetic map, which yielded strong support for the hypothesis that a single whole-genome duplication occurred in the basal vertebrate lineage, preceded and followed by several evolutionarily independent segmental duplications that occurred over chordate evolution.