Despite the rapid growth of recombinant adeno-associated virus (rAAV) gene therapy research, access to high-quality rAAV remains the major limitation to the clinical gene therapy pipeline. Due to the cytoxicity of the Ad helper gene products as well as cytotoxic or cytostatic effects of the AAV Rep proteins, rAAV production is a transient, terminal process. Thus, introducing the essential genetic information into the producer cells remains the major challenge for large-scale rAAV production. The recombinant baculovirus expression vectors (BEVs) and Spodoptera frugiperda (Sf9) cells provides an alternative to transient transfection of human embryonic kidney (HEK) 293 cells. The BEVs replicate and are amplified in the same cell that produces rAAV and the subsequent cell-to-cell spread of the BEVs results in essentially quantitative infection of the culture in any volume bioreactor. In the two BEV system, the AAV vector genome is on one BEV and the second BEV provides the Rep and VP ORFs. In the presence of p5 Rep proteins, the AAV vector genome replicates autonomously from baculovirus. Expression of the Rep and capsid proteins are no longer regulated by the native AAV promoters, therefore, optimal AAV gene expression of the Rep and structural proteins involves engineering the coding sequence to achieve stoichiometric expression of Rep 78, Rep 52, VP1, VP2, and VP3. A systematic approach has been developed to increase the likelihood that the capsid composition is near optimal.