To assess the relationship between novel insulin resistance (IR) indices and the presence and severity of diabetic retinopathy (DR) in patients with type 2 diabetes. This is a cross-sectional study involving 2211 patients. The study outcomes were DR events. The study exposures were IR indices including estimated glucose disposal rate (eGDR), natural logarithm of glucose disposal rate (lnGDR), metabolic insulin resistance score (METS-IR), triglyceride glucose index-body mass index (TyG-BMI), triglyceride glucose index-waist-to-hip ratio (TyG-WHR), and triglyceride/high-density lipoprotein cholesterol(TG/HDL-c ratio). We used binary and multivariate ordered logistic regression models to estimate the association between different IR indices and the presence and severity of DR. Subject work characteristic curves were used to assess the predictive power of different IR indices for DR. DR was present in 25.4% of participants. After adjusting for all covariates, per standard deviation (SD) increases in eGDR (ratio OR 0.38 95% CI 0.32-0.44), lnGDR (0.34 0.27-0.42) were negatively associated with the presence of DR. In contrast, per SD increases in METS-IR (1.97 1.70-2.28), TyG-BMI (1.94 1.68-2.25), TyG-WHR (2.34 2.01-2.72) and TG/HDL-c ratio (1.21 1.08-1.36) were positively associated with the presence of DR. eGDR was strongly associated with severity of DR. Of all variables, eGDR had the strongest diagnostic value for DR (AUC = 0.757). Of the six IR indices, eGDR was significantly associated with the presence and severity of DR in patients with type 2 diabetes. eGDR has a good predictive value for DR. Thus, eGDR maybe a stronger marker of DR.