Although a large percentage of the world population is seropositive for exposure to various strains of adeno-associated virus (AAV), a human parvovirus, AAV has never been identified as an etiologic agent of human disease. Most likely contributing to the pronounced lack of pathogenicity is the fact that AAV is a naturally 'defective' virus that requires a helper virus for productive replication of its genome. Another unusual aspect of wild-type AAV biology is the ability of the virus to establish latent infection by preferential integration of its genome into a specific locus of human chromosome 19. Site-specific integration was a major impetus for the development of recombinant AAV vectors, which typically lack all AAV coding sequences. It was soon realized, however, that expression of at least one species of the virally encoded initiator proteins, Rep78 or Rep68, is necessary for targeted integration of AAV-derived DNA constructs to occur. This article will present a chronological outline of studies characterizing site-specific integration of wild-type AAV sequences and the quasi-random target site selection observed with recombinant AAV vectors.