Chemical neuromodulation of the retina using native neurotransmitters to biomimetically activate target retinal neurons through chemical synapses is a promising biomimetic alternative to electrical stimulation for restoring vision in blindness caused by photoreceptor degenerative diseases. Recent research has shown that subretinal chemical stimulation could be advantageous for treating photoreceptor degenerative diseases but many of the parameters for achieving efficacious chemical neuromodulation are yet to be explored. In this paper, we investigated how the depth at which neurotransmitter is injected subretinally affects the success rate, spike rate characteristics (i.e., amplitude, response latency, and time width), and spatial resolution of chemical stimulation in wild-type Long Evans and photoreceptor degenerated S334ter-3 transgenic rat retinas in vitro. We compared the responses to injections of glutamate at the subretinal surface and two subsurface depths near the outer and inner plexiform layers and found that while injections at all depths elicited robust retinal ganglion cell responses, they differed significantly in terms of the spike rate characteristics and spatial resolutions across injection depths. Shallow subsurface injections near the outer plexiform layer evoked the highest spike rate amplitudes and had the highest spatial resolution and success rates, while deep subsurface injections near the inner plexiform layer elicited the shortest latencies and narrowest time widths. Our results suggest that surface injections are suboptimal for subretinal chemical neuromodulation, while shallow subsurface and deep subsurface injections may optimize high spatial and high temporal resolution, respectively. These findings have great significance for the design and development of a potential neurotransmitter-based subretinal prosthesis.