Gene therapy vehicles must be engineered to overcome numerous barriers that limit delivery efficiency. These barriers arise at every step of the delivery process, including the transit of the vector from injection to a cell surface, receptor binding and uptake, intracellular trafficking, and nuclear entry. The gene transfer properties of the highly promising adeno-associated viral (AAV) vector at each step are determined by its capsid structure. Previous capsid modifications that alter AAV tropism, as well as the existence of multiple AAV serotypes, suggest that the AAV capsid is reasonably plastic. We have taken advantage of this remarkable capsid plasticity to generate a large mutant AAV library (1e6) and select for mutant AAV virons that can overcome several barriers to infection. Specifically, we have selected AAV2 library for infectious particles with altered heparan sulfate (HS) affinity and for the ability to evade an AAV2 immune response. We have generated mutants with lower and higher affinity to heparin, which could prove valuable in controlling the therapeutic zone of an AAV vector in tissues where ECM HS hinders AAV2 diffusion. Furthermore, we have generated vector variants that have resistance to human serum that neutralizes wild type AAV2, yet retain AAV2 gene delivery efficiency. These vectors may enable high gene delivery efficiency even in patients with preexisting immunity, and the locations of point mutations on the capsid surface suggest new regions of functional importance to the virus. These AAV libraries therefore both provide useful variants for gene therapy application and offer a means to dissect AAV biology.