Based on studies in mice, hemophilic dogs, and non-human primates demonstrating long-term (>5 yrs) expression of Factor IX (FIX) after infusion of an AAV vector expressing FIX into the portal vein or the hepatic artery, we undertook a Phase I dose escalation study of AAV-FIX in humans with severe hemophilia B. The first two doses, 2x1011 vg/kg, and 1x1012 vg/kg, were safe but subtherapeutic. Two subjects treated at a dose of 5x1012 vg/kg showed detectable circulating levels of FIX (up to 11.8% and 3% respectively), but expression was transient and accompanied in one case (subject E) by a transient asymptomatic transaminitis. There was never evidence of a FIX inhibitor. Two differences between the large animal models and humans with the disease were hypothesized to contribute to the difference in duration of expression; long-term in hemophilic dogs, short-term in hemophilic humans. First was pre-existing immunity to wild-type AAV-2, which infects humans, but not dogs; and the other was prior exposure to viral hepatitis , found in humans but not in animals. To further assess the roles of viral hepatitis and of the immune response to AAV-2, we treated an additional subject (subject G) at a dose of 1x1012 vg/kg. This subject was 20 yrs. of age and had never been infected with hepatitis. Nevertheless, his transaminases began to rise 3 weeks after vector injection, peaked 6 weeks after injection, and resolved spontaneously as had been seen in subject E. In subject G, magnitude of the peak ALT response was 5-fold less than that found in subject E (5-fold higher dose). Both subjects had similar and low baseline anti-AAV antibody titers. Immune response to AAV-2 was assessed by ELISpot at serial time points before and after vector injection in subject G. The subject’s PBMCs were incubated with a peptide library arrayed in a matrix of 24 pools, each containing 12 peptides of 15-mers overlapping by 10 and spanning the entire VP-1 protein. There was no detectable IFN- γ secretion in response to AAV-2 peptides at baseline, although there was a strong IFN- γ response to PHA. Two weeks after vector infusion, three pools elicited IFN- γ secretion from the subject’s PBMCs. Response to the same pools of peptides, but not to other pools, was repeatedly detected over the next 6 weeks. By week 12, IFN- γ responses were no longer detectable. The matrix array allowed identification of two specific AAV-2 capsid peptides as the T cell immunoreactive epitopes. These peptides are highly conserved in AAV serotypes 1–8. Similar experiments were conducted with a FIX peptide library and demonstrated no response. These data are consistent with a model in which a T cell response to AAV capsid epitopes results in elimination of the transduced cells. This response is only briefly detectable in PBMCs, and hepatitis is not an important risk factor. These immune responses may limit use of standard serotypes of AAV for gene transfer into human liver. Transient immunomodulation may prevent these responses.