Simian immunodeficiency virus (SIV) naturally infects non-human primates in Africa. To date, 40 SIVs have been described both in natural hosts and in heterologous species. These viruses are highly diverse and the majority cluster in 6 relatively equidistant phylogenetic lineages. At least 8 SIVs are currently considered as recombinant viruses, based on different clustering patterns in different genomic regions. Only three types of genomes are known, based on the number of accessory genes: vpr-containing genomes, vpr-vpx containing genomes and vpr-vpu-containing genomes. vpx resulted by a duplication of the vpr gene following non-homologous recombination and is characteristic of SIVs infecting the Papionini tribe of monkeys and HIV-2 in humans. vpu is characteristic of SIVcpz and HIV-1 and may have originated from a recombination involving SIVs from cercopitecini monkeys. SIV seems to be non-pathogenic in the vast majority of natural hosts in spite of a high levels of viral replication. This is probably a consequence of virus-host adaptation, in which the incubation period of the disease generally exceeds the life span of the African primate host. SIVs also have a high propensity for cross-species transmission. In the new host, the outcome may vary from inapparent infection to highly pathogenic, the former being reported for African monkeys, whereas the latter being observed in macaques and humans. The high diversity of SIVs was generated by a high mutation rate due to a low fidelity of the reverse-transcriptase and active viral and host cell turnover, host-dependent evolution and recombination. Cross-species transmission is not rare, however preferential host switching may drive the majority of cross-species transmissions. Numerous SIVs tested so far are able to grow in vitro on human PBMC, therefore it has been postulated that SIV represents a threat for infection of humans in Central Africa and that AIDS is a zoonosis. However, although the simian origin of the two HIV types is broadly acknowledged, there are no data that AIDS is acquired like a zoonosis. SIV may undergo adaptation in the new human host in order to emerge in the general population. The study of SIV in their natural hosts should provide important clues to the real threat to human populations and also elucidate the mechanisms associated with a long-term persistent viral infection without clinical consequences for the host.