Information that defines the depth of activation of retinal neurons is useful in considering strategies for stimulation with a retinal prosthesis, or interpreting the results from human studies that have previously been performed. The purpose of this study was to test the assertion that electrical pulse durations >0.5 msec preferentially stimulate retinal neurons deep to the ganglion cell layer. Thirteen Dutch-belted rabbits (1.2–2.0 kg) were used in this study. A Goldmann-like dome was used to deliver photic stimuli to the retina to measure the electroretinogram (ERG) and the light-induced cortical potential (VECP). Then, a micromanipulator was used to position a 500 μm inner diameter bipolar electrode near the visual streak on the epi-retinal surface. Symmetric biphasic pulses (7–1600 μA; 0.25 msec and 2.0 msec pulses per phase; biphasic pulses delivered at 2 Hz) were delivered to the retina with a current source. Extra-dural electrodes were used to record electrical evoked cortical potentials (EECPs) over the occipital cortex by performing 50 consecutive computer-averaged stimulations. The effect on the EECP of sequential epi-retinal (i.e. return electrode on epi-retinal surface) vs. trans-retinal (i.e. return electrode behind sclera) stimulation was compared. The effect upon the ERG, VECP and EECP was then assessed after 2,3,dihydroxy-6-nitro-7-sulfamoyl-benzo-ƒ-quinoxaline (NBQX) at 112 μM concentration, d-2-amino-7-phosphonoheptanoic acid (D-AP7) at 1200 μM concentration, and l-amino-4-phosphonobutyrate (APB) at 300 μM concentration were delivered into the vitreous cavity to selectively block neuronal input to the retinal ganglion cells. Median values were reported. The amplitudes of the light-induced ERG and VECP were markedly reduced by instillation of the intra-vitreal synaptic blocking drugs. By comparison, pharmacological blockade of input to the retinal ganglion cells did not significantly alter the threshold charge or amplitude of the electrically-induced cortical responses (P > 0.05). For the electrical stimuli, there was no significant difference in threshold charge for the EECP for epi-retinal vs. trans-retinal stimulation (P > 0.05). The amplitude of the EECP increased linearly with increasing charge using both 0.25 msec and 2.0 msec pulses, even after synaptic blockade of input to the ganglion cells. The lack of obvious degradation of cortical amplitudes after drug instillation indicates that neurons of the middle retina are not being preferentially driven with epi-retinal stimulation, at least not with stimulus pulses up to 2.0 msec in duration. This conclusion is in contrast to prior evidence that 2.0 msec pulses would preferentially stimulate deeper retinal neurons, specifically the bipolar cells. Based upon our own observations in other studies, we believe that preferential stimulation of the middle retina in fact can be achieved by epi-retinal stimulation, by using pulse durations at least 5 times longer than those used in this study.