Computed tomographic angiography (CTA) is considered to have limited accuracy for quantifying exact percent diameter stenosis in coronary arteries. However, most studies evaluating CTA use quantitative coronary angiography (QCA) as the gold standard, a technique with its own limitations. We sought to determine whether CTA measurements of stenosis severity correlate better with intravascular ultrasound (IVUS) than with QCA. Luminal dimensions of 67 de novo coronary lesions were measured by CTA, IVUS, and QCA. IVUS was performed when lesion severity by angiography was equivocal. Mean percent diameter stenosis by QCA was 51 +/- 9.8% and mean IVUS minimal luminal area was 3.8 +/- 1.8 mm(2). There was a moderate correlation between CTA minimal luminal area and IVUS minimal luminal area (r(2) = 0.41, p <0.001), but no relation between CTA and QCA measurements of minimal luminal diameter (r(2) = 0.01, p = 0.57) or diameter stenosis (r(2) = 0.02, p = 0.31). There was also no relation between IVUS minimal luminal area and QCA diameter stenosis (r(2) = 0.01, p = 0.50). When lesions with moderate or severe calcification were excluded, the correlation between CTA minimal luminal area and IVUS minimal luminal area was good (r(2) = 0.68, p <0.001). In conclusion, in this cohort of patients with intermediate-grade lesions on cardiac catheterization, absolute measurements of stenosis severity on CTA correlated with IVUS but not with QCA. Our findings suggest that limitations of quantitative coronary angiography as a gold standard need to be considered in studies evaluating the accuracy of coronary CTA.