Previous research has suggested that human herpesvirus-6 (HHV-6) may integrate into host cell chromosomes and be vertically transmitted in the germ line, but the evidence--primarily fluorescence in situ hybridization (FISH)--is indirect. We sought, first, to definitively test these two hypotheses. Peripheral blood mononuclear cells (PBMCs) were isolated from families in which several members, including at least one parent and child, had unusually high copy numbers of HHV-6 DNA per milliliter of blood. FISH confirmed that HHV-6 DNA colocalized with telomeric regions of one allele on chromosomes 17p13.3, 18q23, and 22q13.3, and that the integration site was identical among members of the same family. Integration of the HHV-6 genome into TTAGGG telomere repeats was confirmed by additional methods and sequencing of the integration site. Partial sequencing of the viral genome identified the same integrated HHV-6A strain within members of families, confirming vertical transmission of the viral genome. We next asked whether HHV-6A infection of naïve cell lines could lead to integration. Following infection of naïve Jjhan and HEK-293 cell lines by HHV-6, the virus integrated into telomeres. Reactivation of integrated HHV-6A virus from individuals' PBMCs as well as cell lines was successfully accomplished by compounds known to induce latent herpesvirus replication. Finally, no circular episomal forms were detected even by PCR. Taken together, the data suggest that HHV-6 is unique among human herpesviruses: it specifically and efficiently integrates into telomeres of chromosomes during latency rather than forming episomes, and the integrated viral genome is capable of producing virions.