Objective: To investigate the early identification of proliferative diabetic retinopathy (PDR) to coronary atheroslerotic heart disease (CHD). Methods: A total of 351 patients with type 2 diabetes mellitus (T2DM) individuals [211 non-diabetic retinopathy (NDR) and 140 proliferative diabetes mellitus (PDR)] were recruited from one of the national diabetes mellitus center from China. CHD was newly diagnosed and defined by the evidences of coronary angiography or computed tomography coronary angiography. The area under curve (AUC) was performed. Discrimination and reclassification were also used to evaluate the value of addition of PDR to the base model. Results: In our present study, with the model estimating risk of asymptomatic CHD based on the United Kingdom Prospective Diabetes Study (UKPDS) engine, we found that the level of AUC was greatly improved from 0.583 (95% CI: 0.51–0.66) to 0.697 (95% CI: 0.641–0.752) between the models with PDR and without PDR. Besides, the C statistic was increased from 0.746 (95% CI: 0.681–0.811) to 0.762 (95% CI: 0.699–0.825; P=0.73) by the addition of PDR to a base model to identify asymptomatic CHD. There was a continuous net reclassification indexes (NRI) of 5.9% (95% CI: 1.4%–10.2%, P=0.017) for the asymptomatic CHD. The absolute integrated discrimination index (IDI) was 0.004 (P=0.02). Conclusion: Compared with the traditional risk factors, our results improved the identification of asymptomatic CHD, which further confirmed that PDR could identify cardiovascular risk stratification in Chinese diabetic individuals.