Purpose To test how the PhNR of the flash ERG is affected in human nonproliferative diabetic retinopathy (NPDR). Methods The PhNR was elicited with red stimuli (5 cd s/m2 with 4-min duration) and blue background (10 cd/m2). Standard Ganzfeld flash ERGs were recorded according to the ISCEV standard for the clinical electroretinogram (2004). A total of 81 diabetic patients with different severity levels of NPDR were examined. Forty-three age-matched normal controls were also studied. Results The amplitude of PhNR decreased significantly as DR progressed, while the implicit time was prolonged. Amplitudes of the PhNR in the control group, no DR, mild NPDR, moderate NPDR, and severe NPDR were 78.1 ± 15.1 μV, 69.0 ± 17.8 μV, 64.5 ± 13.2 μV, 45.9 ± 9.0 μV, and 33.7 ± 10.8 μV respectively, and the implicit times of PhNR were 71.5 ± 5.0, 72.0 ± 6.2, 73.6 ± 5.0, 75.7 ± 6.1, and 82.9 ± 7.8 min respectively. Compared to the control group, the reduction of PhNR amplitude in all diabetic groups was statistically significant. However, except for the OPs, the percent reduction of the amplitude of standard ERG waves was far less than that of the PhNR. The percent decrease in amplitude of the PhNR and ∑OPs was not significantly different in any diabetic group. However, the reduction of the amplitude of ∑OPs (and other standard ERG waves) was statistically significant only in moderate and severe NPDR groups. Conclusions PhNR is a sensitive indicator of the function of inner retina in diabetic patients. There is a potential role for the PhNR in assessing inner retinal damage and evaluating the effect of treatment in NPDR.