Background: Iterative reconstruction (IR) in cardiac CT has been shown to improve confidence of interpretation of noninvasive coronary CT angiography (CTA). Objective: We hypothesized that IR would also improve the quality of vasodilator stress coronary CT images acquired with low tube voltage to assess myocardial perfusion and the accuracy of the detection of perfusion abnormalities by using quantitative 3-dimensional (3D) analysis. Methods: We studied 39 consecutive patients referred for coronary CTA (256-slice scanner; Philips), who underwent additional imaging (100 kV, prospective gating) with regadenoson (0.4 mg; Astellas). Stress images were reconstructed with different algorithms: filtered back projection (FBP) and IR (iDose; Philips). Image quality was quantified by signal-to-noise and contrast-to-noise ratios in the blood pool and the myocardium. Then, FBP and separately IR images were analyzed with custom 3D analysis software to quantitatively detect perfusion defects. Accuracy of detection was compared with perfusion abnormalities predicted by coronary stenosis >50% on coronary CTA. Results: Five patients with image artifacts were excluded. In the remaining 34 patients, both signal-to-noise and contrast-to-noise ratios increased with IR, indicating improvement in image quality compared with FBP. For 3D perfusion analysis, 10 patients with normal coronary arteries were used as a reference to correct for x-ray attenuation variations in normal myocardium. In the remaining 24 patients, reduced noise levels in the IR images compared with FBP resulted in tighter attenuation distribution and improved detection of perfusion abnormalities. Conclusion: IR significantly improves image quality on regadenoson stress CT images acquired with low tube voltage, leading to improved 3D quantitative evaluation of myocardial perfusion.