Recombinant adeno-associated virus vectors (rAAV) show promise in preclinical trials for the treatment of genetic diseases including hemophilia. Liver-directed gene transfer results in a slow rise in transgene expression, reaching steady-state levels over a period of 5 weeks concomitant with the conversion of the single-stranded rAAV molecules into high-molecular-weight concatemers in about 5% of hepatocytes. Immunohistochemistry and RNA in situ hybridization show that the transgene product is made in about approximately 5% of hepatocytes, suggesting that most rAAV-mediated gene expression occurs in hepatocytes containing the double-stranded concatemers. In this study, the mechanism(s) involved in stable transduction in vivo was evaluated. While only approximately 5% of hepatocytes are stably transduced, in situ hybridization experiments demonstrated that the vast majority of the hepatocytes take up AAV-DNA genomes after portal vein infusion of the vector. Two different vectors were infused together or staggered by 1, 3, or 5 weeks, and two-color fluorescent in situ hybridization and molecular analyses were performed 5 weeks after the infusion of the second vector. These experiments revealed that a small but changing subpopulation of hepatocytes were permissive to stable transduction. Furthermore, in animals that received a single infusion of two vectors, about one-third of the transduced cells contained heteroconcatemers, suggesting that dimer formation was a critical event in the process of concatemer formation. To determine if the progression through the cell cycle was important for rAAV transduction, animals were continuously infused with 5'-bromo-2'-deoxyuridine (BrdU), starting at the time of administration of a rAAV vector that expressed cytoplasmic beta-galactosidase. Colabeling for beta-galactosidase and BrdU revealed that there was no preference for transduction of cycling cells. This was further confirmed by demonstrating no increase in rAAV transduction efficiencies in animals whose livers were induced to cycle at the time of or after vector administration. Taken together, our studies suggest that while virtually all hepatocytes take up vector, unknown cellular factors are required for stable transduction, and that dimer formation is a critical event in the transduction pathway. These studies have important implications for understanding the mechanism of integration and may be useful for improving liver gene transfer in vivo.