Adeno-associated virus (AAV) was discovered about 30 yr ago as a contaminant of adenovirus preparations. Since its discovery, researchers have described many unique characteristics of AAV biology that have made it attractive as a potential vector for gene therapy. For example, AAV is not pathogenic, approx 80% of adults in the United States are seropositive, but in no case has the virus been implicated as the etiological agent for a human disease. AAV is a defective parvovirus with a single-stranded DNA genome of 4.6 kb comprising two open reading frames coding for nonstructural (Rep) and structural (Cap) proteins. The entire genome is flanked by two identical 145-basepair (bp) inverted terminal repeats (ITR), as shown in Fig. 1. Possibly, the most striking property of the virus is its inability to productively infect healthy cells in the absence of a coinfection by a helper virus (either adenovirus or herpesvirus). Thus, AAV vectors are inherently replication defective. Infection in the absence of helper virus leads to integration of the AAV genome at a specific site on the q arm of chromosome 19 to establish latent infection (1-3). This is the only known example of site-specific integration by a mammalian virus and suggests the possibility of a Fig. 1. Structure of AAV2 genome. Black boxes represent the inverted terminal repeats, arrows designate the three viral promoters at map positions 5, 19, 40, and open boxes denote coding regions. The left open reading frame (ORF) encodes four Rep proteins, Rep78, Rep68, Rep52, and Rep40. The right ORF encodes three Cap proteins, VP1, VP2, and VP3.