Phase 1 and phase 2 gene therapy trials using intramuscular (IM) administration of a recombinant adeno-associated virus serotype 1 (rAAV1) for replacement of serum alpha-1 antitrypsin (AAT) deficiency have shown long-term (5-year) stable transgene expression at approximately 2% to 3% of therapeutic levels, arguing for the long-term viability of this approach to gene replacement of secreted serum protein deficiencies. However, achieving these levels required 100 IM injections to deliver 135 mL of vector, and further dose escalation is limited by the scalability of direct IM injection. To further advance the dose escalation, we sought to bridge the rAAV-AAT clinical development program to regional limb perfusion, comparing two methods previously established for gene therapy, peripheral venous limb perfusion (VLP) and an intra-arterial push and dwell (IAPD) using rAAV1 and rAAV8 in a non-human primate (rhesus macaque) study. The rhesus AAT transgene was used with a c-myc tag to enable quantification of transgene expression. 5 cohorts of animals were treated with rAAV1-IM, rAAV1-VLP, rAAV1-IAPD, rAAV8-VLP, and rAAV8-IAPD (n = 2-3), with a dose of 6 × 1012 vg/kg. All methods were well tolerated clinically. Potency, as determined by serum levels of AAT, of rAAV1 by the VLP method was twice that observed with direct IM injection; 90 μg/mL with VLP versus 38 μg/mL with direct IM injection. There was an approximately 25-fold advantage in estimated vector genomes retained within the muscle tissue with VLP and a 5-fold improvement in the ratio of total vector genomes retained within muscle as compared with liver. The other methods were intermediate in the potency and retention of vector genomes. Examination of muscle enzyme (CK) levels indicated rAAV1-VLP to be equally safe as compared with IM injection, while the IAPD method showed significant CK elevation. Overall, rAAV1-VLP demonstrates higher potency per vector genome injected and a greater total vector retention within the muscle, as compared to IM injection, while enabling a much greater total dose to be delivered, with equivalent safety. These data provide the basis for continuation of the dose escalation of the rAAV1-AAT program in patients and bode well for rAAV-VLP as a platform for replacement of secreted proteins.