Background: Our goal was to examine the causal link between blood metabolites, their ratios, and the risk of developing proliferative diabetic retinopathy (PDR) from a genetic insight.Methods: Summary‐level data about 1400 blood metabolites and their ratios, as well as PDR, were sourced from prior genome‐wide association studies (GWAS). A two‐sample univariate and multivariate Mendelian randomization (MR) approach was utilized. Additionally, metabolic pathway analysis and sensitivity analysis were also conducted.Results: After adjusting for multiple tests, four blood metabolites significantly correlated with PDR risk. Two ceramides, including glycosyl‐N‐palmitoyl‐sphingosine (d18:1/16:0) (odds ratio [OR] = 1.12, 95% confidence interval (CI): 1.06–1.17, p < 0.001, false discovery rate (FDR) = 0.005) and glycosyl‐N‐behenoyl‐sphingadienine (d18:2/22:0) (OR = 1.11, 95% CI: 1.06–1.16, p < 0.001, FDR = 0.017), were linked to increased risk. Additionally, 3‐methylcytidine (OR = 1.05, 95% CI: 1.03–1.08, p < 0.001, FDR = 0.021) also posed a risk, whereas (N(1)+N(8))‐acetylspermidine (OR = 0.91, 95% CI: 0.87–0.94, p < 0.001, FDR = 0.002) appeared protective. Multivariable MR analysis further confirmed a direct, protective effect of (N(1)+N(8))‐acetylspermidine on PDR risk (OR = 0.94, 95% CI: 0.89–1.00, p = 0.040). The sensitivity analysis results indicated that evidence for heterogeneity and pleiotropy was absent.Conclusion: These metabolites have the potential to be used as biomarkers and are promising for future research into the mechanisms and drug targets for PDR.