We assessed the diagnostic performance of stress- and rest-dual-energy computed tomography (DECT) and their incremental value when used with coronary CT angiography (CCTA) compared with combined invasive coronary angiography (ICA)/cardiovascular magnetic resonance (CMR) for detecting hemodynamically significant stenosis causing a myocardial perfusion defect. Forty patients (30 men; mean age, 63.4 ± 8.8 years) with known or suspected coronary artery disease detected by CCTA underwent stress- and rest-DECT, CMR, and ICA. DECT iodine maps were compared with CMR on a per-segment and per-vessel basis. Diagnostic value of CCTA was assessed on a per-vessel basis before and after stress- and rest-DECT and compared to that of ICA/CMR. Compared to CMR, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of rest-DECT for detecting segment (vessel)-based perfusion defects were 29 % (46 %), 88 % (79 %), 56 % (61 %), and 70 % (67 %), respectively. Corresponding values using stress-DECT were 73 % (94 %), 83 % (74 %), 70 % (72 %), and 85 % (95 %), respectively. There was fair (κ = 0.39) agreement between rest- and stress-DECT iodine maps in identifying segments with perfusion defects. Compared with the ICA/CMR for identifying hemodynamically significant stenoses, per-vessel territory sensitivity, specificity, PPV, and NPV of CCTA were 91, 56, 55, and 91 %, respectively; those using CCTA/rest-DECT were 42, 83, 59, and 70 %, respectively; and those using CCTA/stress-DECT were 87, 79, 71, and 91 %, respectively. The area under the receiver operating characteristic curve decreased from 0.74 to 0.62 (P = 0.06) using CCTA/rest-DECT but increased to 0.83 (P = 0.02) using CCTA/stress-DECT. Stress-DECT has incremental value when used with CCTA for detecting hemodynamically significant stenoses.