To explore serum zinc-α2-glycoprotein (ZAG) changes in patients with or without premature coronary artery disease (PCAD) and its association with several cardiovascular risk factors. A total of 3,364 patients who were undergone coronary angiography in Peking Union Medical College Hospital were screened. According to the degree of coronary artery stenosis, the number of 364 patients with PCAD (age <55 years in males and <65 years in females) and 126 age and gender matched patients without premature coronary artery disease (NPCAD) were recruited in our present study. In addition, 182 age and gender matched healthy controls were also enrolled. Serum ZAG levels were determined by enzyme-linked immunosorbent assay (ELISA) method. Serum ZAG were significantly lower in the PCAD (8.03 ± 1.01 vs. 8.78 ± 1.89 μg/mL, < 0.05) and NPCAD groups (8.28 ± 1.61 vs. 8.78 ± 1.89 μg/mL, < 0.05), respectively, when compared with the controls. Multiple regression analysis showed that PCAD was independently associated with serum ZAG levels ( = -0.289, = 0.002). The probability of PCAD in subjects with low tertile ZAG levels was 2.48-fold higher than those with high tertile levels after adjusting for other confounders [OR = 3.476, 95% CI 1.387-8.711, = 0.008]. This phenomenon was more likely to be observed in male subjects with BMI <24 kg/m. The receiver operating curve (ROC) analysis showed a weak diagnostic performance of serum ZAG for PCAD (AUC = 0.659, 95% CI 0.612-0.705, < 0.05). At the cutoff value of 7.955 μg/mL serum ZAG, the sensitivity and specificity for differentiating patients with PCAD from controls were 50.5 and 78.0%, respectively. The combination of ZAG with other clinical variables including age, gender, BMI, SBP, FBG, TC, HDL-C, Cr, and Urea had significantly improved the diagnosis accuracy with a sensitivity of 82.6%, a specificity of 95.0%, and AUC of 0.957 (95% CI, 0.940-0.975, < 0.05). Serum ZAG levels were firstly found to be decreased in Chinese PCAD patients. Subjects with lower ZAG levels were more likely to have PCAD, especially for male subjects with BMI <24 kg/m. ZAG might be the potential diagnostic biomarkers for PCAD patients, and the combination of ZAG and clinical variables had higher discriminative performance.