Introduction. Coronary Artery Disease (CAD) is one of the disease entity that leading cause of morbidity and mortality in worldwide. Treadmill test is part of the diagnostic modality which readily available to assess possibility of narrowing coronary artery and guiding us whether we need for the further investigation. Despite of that, treadmill test has limitation in diagnostic accuracy. Duke Treadmill Score (DTS) was also tested as a diagnostic score, and shown to predict significant CAD better than the ST-segment response alone. Methods. This is a cross-sectional study performed in adult patients with stable CAD that underwent treadmill test and coronary angiography in outward patient clinic of the Integrated Cardiac Service in Cipto Mangunkusumo Hospital between January 2011 and December 2013. Results. A total of 103 patients in this study, thirty nine patients (37,9 %) had significant CAD in coronary angiography. Briefly, mean age was 54,71 years and 55 patients (53,4 %) were females. The most common CAD risk factor was hypertension (51,5 %). A mean of DTS score was -3.53, which mostly categorized as intermediate risk (89,3 %). Based on DTS results, cut-off point was determined by using Receiver Operator Curve (ROC) method, in which value of -8,85 considering as a cut-off point. Sensitivity and specificity value of DTS were 28 % (CI 95 %: 17 % to 44 %), and 95 % (CI 95 %: 87 % to 98 %). Positive and negative predictive value were 79 % (CI 95 %: 52 % to 92 %) and 69 % (CI 95 %: 58 % to 77 %). Positive and negative likelihood ratio were 6.02 and 0.75. Conclusions. DTS has a good performance in predicting significant CAD at cut-off point -8,85 in patients with positive treadmill test.