The aim of this study was to define the optimal cutoff value of the neutrophil/lymphocyte ratio (NLR) to predict severe grades of diabetic retinopathy (DR). Methods: A total of 40 patients with proliferative DR (PDR) and 40 patients with severe non-proliferative DR (NPDR) were included this prospective, case control study, and 35 age-and sex-matched healthy subjects were recruited as a control group. White blood cell (WBC) count mean values and ratios were compared between the groups. Results: The groups were statistically similar in terms of age and sex. The neutrophil, lymphocyte, monocyte, platelet counts, and main platelet volume (MPV) values were similar in all 3 groups (all p values >0.05). The mean NLR was 2.67±1.02 in the PDR cases, 2.16±0.58 in severe NPDR, and 1.85±0.49 in the control group, which represented a statistically significant difference between the 3 groups (p=0.003). In post-hoc analysis, the NLR of the PDR and severe NPDR groups was statistically significantly greater than that of the control group (p=0.002 and p=0.048, respectively), but there was no statistically significant difference between the PDR and severe NPDR groups (p=0.083). The monocyte/lymphocyte, platelet/lymphocyte, and MPV/lymphocyte ratios were also similar in all 3 groups (all p values >0.05). Conclusion: An NLR value of 2.11 or more predicted DR (PDR or severe NPDR) with a sensitivity of 76% and a specificity of 80%.