Objective: To compare retinal function assessed by full-field electroretinography (ffERG) and multifocal electroretinography (mfERG) in diabetes without retinopathy, diabetes with moderate non-proliferative diabetic retinopathy (NPDR) and in the absence of diabetes. Methods: Scotopic and photopic ffERG and mfERG was made in non-fasting volunteers, including 26 diabetic participants without retinopathy, 22 diabetic participants with moderate NPDR and 22 participants without diabetes using full International Society for Clinical Electrophysiology of Vision protocols. Results: Of the ffERG responses, significant deviation (p ≤ 0.05, corrected for multiple sampling and other relevant confounders) from the non-diabetic participants was seen in the diabetic participants only for the OP1-OP3 oscillatory amplitudes and the OP2 implicit time. This finding was independent of whether retinopathy was present or not. For the mfERG, minor amplitude or implicit time deviations were found for a small number of rings (R2, R4 and R5). Receiver of operating characteristic analysis showed that the single most prominent abnormality of the ffERG in diabetes, regardless of whether retinopathy was present or not, was the OP2 implicit time (area under the curve ≥ 0.80). Conclusion: This bi-modal study of electroretinographic characteristics found that the most prominent anomaly associated with diabetes was a prolongation of the implicit time of the OP2 of the scotopic ffERG, while the most prominent added effect of non-proliferative diabetic retinopathy was a further prolongation of the OP2 implicit time. Although the variation in ERG characteristics is far too large for diagnostic purposes, the close association of the oscillatory potentials with the amacrine cells of the retina indicate that their function is particularly sensitive to diabetes.