Objective：Coronary slow flow phenomenon（CSFP）is a pathological condition characterized by delayed coronary blood flow in the absence of significant stenosis on coronary angiography，which predisposes patients to angina pectoris and cardiovascular events. Current diagnosis relies heavily on invasive investigations，and simple，effective non-invasive predictive tools are lacking. This study aimed to investigate the expression levels of serum growth differentiation factor-15（GDF-15）in CSFP and evaluate its predictive value，with the goal of providing an alternative to the current diagnostic paradigm dependent on invasive coronary angiography（the gold standard）. Methods：Patients undergoing coronary angiography at the Affiliated Hospital of Qingdao University were enrolled between December 2023 to June 2024. Those with angiographically normal coronary arteries were assigned to the normal group（n=42），while patients exhibiting delayed coronary flow without significant stenosis comprised the CSFP group（n=69）. The serum GDF - 15 levels from the day before the surgery were measured using enzyme-linked immunosorbent assay（ELISA）. Logistic regression was performed and collinearity was examined；diagnostic efficacy was assessed by using receiver operating characteristic（ROC）curves and crossvalidated with Bootstrap internal validation. Results：GDF - 15 levels were significantly higher in the CSFP group versus controls [957.01（716.27，1 373.16）ng/L vs. 745.14（585.43，812.41）ng/L；z=- 4.14，P ＜ 0.001]. Both univariate and multivariate logistic regression adjusted for body mass index（BMI）and other confounders showed that each 1-unit increase in ln（GDF-15）（corresponding to a 2.718-fold increase in raw concentration）was associated with a 14.06-fold higher CSFP risk（95% CI：1.82-68.76，P ＜ 0.05）. Conversely，for each 1 mmol/L increase in high-density lipoprotein cholesterol（HDL-C），the risk was reduced by 16%（OR=0.16， 95%CI：0.07-0.34，P ＜ 0.05）. ROC analysis indicated that GDF-15 alone had an AUC of 0.791 for diagnosing CSFP. Combining GDF -15 with HDL -C increased the AUC to 0.953，improving sensitivity from 57.97% to 91.30%；furthermore，cross -validation and Bootstrap indicated cut-off generally stable. Conclusion：Elevated serum GDF-15 level in CSFP patients establishes its potential as a non-invasive early warning biomarker. The combined GDF-15/HDL-C diagnostic model demonstrates substantially improved accuracy， suggesting its utility as a practical clinical screening tool. This approach could reduce dependence on invasive coronary angiography for CSFP detection.