Introduction For gene therapy applications in the primate retina, intravitreal injections (IVT) of adeno-associated virus (AAV) have been shown to target the retina inefficiently, mainly in retinal ganglion cells within a small perifoveal region. The internal limiting membrane (ILM) has been previously hypothesized to be a barrier to broader transduction. We further hypothesized that the local concentration and residence time of AAV at the vitreo-retinal interface was insufficient for efficient retinal transduction. Here, we investigate gene delivery to the retina in rhesus macaque with regular IVT injection as compared to a novel surgical technique “Peel and Puddle” aimed at overcoming these potential barriers. Methods Three eyes of rhesus macaques were injected with AAV-CMV-GFP via a traditional IVT at a concentration of 1E+11 particles (2/2 or Anc80 capsid). Another two eyes from different rhesus macaques had the ILM surgically removed followed by the concentrated application of AAV2 onto the peeled area, known as “Peel and Puddle” (PP), also at 1E+11 particles. The PP procedure consisted of a vitrectomy, posterior hyaloid separation, ILM staining and subsequent ILM peel over the majority of the macula. This was followed by fluid-air exchange, and deposition of the virus on the peeled area with 30 minutes of supine positioning generating a “puddle” onto the retina aimed at maximizing the AAV residence time at an undiluted concentration. Results GFP expression in the PP condition was dramatically greater than with a regular intravitreal injection, both in intensity and the region of transduction. In the PP eyes, fluorescence was noted in an area that matched the extent of the ILM peel and puddle, as shown by imaging and histology whereas traditional IVT demonstrated the expected parafoveal dim pattern of transduction. All eyes, irrespective of the type of IVT injection, presented clinically significant signs of inflammation, which was managed effectively by intravitreal steroid injections. Histological sections show increase microglia infiltration in both groups compared to un-injected controls. Conclusions The vitreous, ILM, and/or physical parameters of the vector delivery significantly limit gene transfer to the retina. Surgical minimization of these barriers demonstrated highly increased levels and distribution of transduction. Inflammation following intravitreal injection with a non-self transgene is of concern and requires further study.