Background The role of chronic hyperglycaemia as a coronary artery disease (CAD) risk factor is well-known, and the glycemic variability is still a matter of debate. The aim of this study was to investigate the association of admission glycemic excursion and hemoglobin A(1c) (HbA(1c)) with the presence and severity of CAD in patients with undiagnosed diabetes mellitus (DM). Methods We studied 286 newly diagnosed DM patients without prior revascularization undergoing coronary angiography for suspected ischaemic chest pain. Patients were grouped into those with CAD and without CAD according to angiographic results. The severity of CAD was assessed using the Gensini score. Glycemic variability, indicated as the mean amplitude of glycemic excursions (MAGE), was determined by a continuous glucose monitoring system. Serum levels of HbA(1c) and high-sensitive C-reactive protein (hs-CRP) as well as plasma concentrations of fasting glucose, lipids and creatinine were measured in all patients. Predictors of CAD were determined using multivariate Logistic regression model and receiver-operating characteristic (ROC) curves. Results The newly diagnosed DM patients with CAD were older, and more were male and current cigarette smokers compared with the patients without CAD. The CAD group had significantly higher levels of MAGE and HbA(1c). Individuals with high levels of HbA(1c) (>= 7%) or MAGE (>= 3.4 mmol/L) had also significantly higher CAD prevalence. Logistic regression analysis revealed that high MAGE level and high HbA(1c) level were independent predictors for CAD. The area under the receiver-operating characteristic curve for MAGE (0.606, P=0.005) was superior to that for HbA(1c) (0.582, P=0.028). Gensini score closely correlated with age, MAGE, HbA(1c), hs-CRP, creatinine and total cholesterol. Multivariate analysis indicated that age (P <0.001), MAGE (P <0.001), HbA(1c) (P=0.022) and hs-CRP (P=0.005) were independent determinants for Gensini score. Conclusions Both admission glycemic excursion and chronic hyperglycaemia are associated with the severity of CAD in newly diagnosed DM patients. MAGE displays a significant value in predicting CAD in patients with undiagnosed diabetes even more than HbA(1c). Chin Med J 2012;125(1):38-43