Adeno-associated viral (AAV) vectors show great potential for therapeutic gene delivery for monogenic diseases, including hemophilia. Major limitations of this approach are the inability to persist in dividing cells and the restrictive packaging capacity of AAV. Gene targeting, the ability to make site-directed changes to the genome, has been a powerful tool for genetic discovery. Until recently, the low efficiency of targeting has rendered it unlikely to be of therapeutic value for most genetic diseases. Zinc finger nucleases (ZFNs) are engineered proteins capable of site-directed DNA cleavage, known to increase the efficiency of gene targeting. Our lab has previously demonstrated in vivo gene targeting of the liver using AAV to deliver ZFNs and therapeutic donor in neonatal mice, where the rapid proliferation of hepatocytes may promote genome editing through homology directed repair (HDR). It was unknown whether the success of this approach could be replicated in adult mice (with predominantly quiescent hepatocytes, unlikely to be amenable to HDR). We hypothesized that in the absence of HDR, the non-homologous end-joining (NHEJ) pathway may promote gene targeting in adult mice. Indeed, homology independent vector integration was sufficient to drive robust expression of clotting factor IX, deficient in hemophilia B, and correct the disease phenotype in mice treated as adults. This approach was adapted to create a general platform for liver-directed protein replacement therapy. Using ZFNs targeting the mouse albumin locus, we achieved long-term therapeutic expression of human factors VIII and IX in mouse models of hemophilia A and B as well as four different therapeutic enzymes deficient in lysosomal storage disorders. To test our hypothesis that in vivo genome editing relies on different DNA repair mechanisms in neonatal and adult mice, we applied multiple techniques including a novel reporter construct, southern blot, comparisons of donors with and without arms of homology, and a mouse model of NHEJ deficiency. These data indicate HDR is the primary mechanism of genome editing in neonatal mouse livers, and dispensable for therapeutically relevant levels of genome editing in adult mice. These results have implications in designing safer and more efficacious genome editing therapies.