Abstract： Objective To investigate the bidirectional causal relationship between lipid and diabetic retinopathy(DR).Methods Data of genome-wide association study on lipid and DR were collected.The bidirectional Mendelian randomization(MR)of two samples was performed by inverse variance weighted method,MR-Egger regression method and weighted median method.Cochran's Q test was used to check the heterogeneity of the results and the MR-Egger regression method to evaluate of the results.Results The results of inverse variance weighted method showed that phosphatidylethanolamine(16:0_18:2)(OR=0.916,95%CI:0.864～0.971,P=0.003,PFDR=0.174)and triacylglycerol 50:2(OR=0.850,95%CI:0.760～0.952,P=0.005,PFDR=0.174)decreased the risk of DR.The results of reverse MR analysis showed that DR was causally associated with increased diglycerol(16:0_18:2)(OR=1.064,95%CI:1.021～1.109,P=0.003,PFDR=0.194)and triacylglycerol 50:1(OR=1.067,95%CI:1.026～1.109,P=0.001,PFDR=0.194).The results of various MR methods were basically consistent,and the results of sensitivity analysis showed that the above results were consistent and reliable.Conclusion Phosphatidylethanolamine(16:0_18:2)and triacylglycerol 50:2 have negative causal associations with the risk of DR,while DR has positive causal effects on diglycerol(16:0_18:2)and triacylglycerol 50:1.