Abstract: Objective To study the relationship between magnesium level in serum of patients with coronary heart disease and the extent of coronary artery stenosis. Methods Patients were selected from Guangzhou Coronary Artery Disease Cohort in 3 hospitals in Guangzhou from October 2008 to December 2011. Patients' characteristics of smoking, alcohol consumption, physical activity and admission condition were collected by cross-sectional method. Their fasting blood specimens in the second day morning after admission were taken for test of blood routine, serum magnesium, and C-active protein (CRP). Univariate and multivariate logistic regressions were used to analyze the data. Results There were 1 980 patients in the Guangzhou Coronary Artery Disease Cohort, and 979 subjects were included in the final analysis. The average age was (63.63±11.84) years for the total patient cohort and (62.62±11.03) for the subjects included (P<0.01), the average BMI was (23.88±3.33) for the cohort and (23.87±3.36) for the subjects (P>0.05), and proportion of the males was 65.60% for the cohort and 73.24% for the subjects (P<0.01). In the univariate logistic regression, the relationship between different serum magnesium level and coronary artery stenosis >75% had statistically significant in male, but not in female. After adjusted for age, BMI, systolic blood pressure, smoking, alcohol consumption, physical activity in the multivariate logistic regression, the serum magnesium levels at 0.39-0.84, 0.92-0.99 and >0.99 mmol/L had significant risk for coronary artery stenosis >75% compared with those at 0.85-0.91mmol/L in males (OR: 1.93, 1.77, 2.14, respectively). Conclusion Serum magnesium level and coronary artery stenosis in male patients were not a simple liner relationship. It might be conducive to the prevention of coronary heart disease to keep serum magnesium at moderate level.