To prospectively evaluate accuracy of gadolinium-enhanced three-dimensional (3D) magnetic resonance (MR) angiography with gadodiamide and gadopentetate dimeglumine (0.1 mmol/kg), with intraarterial DSA as reference standard, for imaging abdominal and iliac arterial stenoses. The study was approved by all institutional review boards; informed consent was obtained from each subject before procedures. Two hundred forty-seven subjects were included; 240 received either contrast agent and were available for safety analysis; 222 were available for accuracy analysis. Enhanced 3D MR angiography and DSA were performed; image data were evaluated in a double-blinded randomized study. Stenoses were classified as not relevant (<50% stenosis) or relevant (> or =50%). For detection of main stenosis, accuracy with enhanced 3D MR angiography compared with that with DSA was determined. The difference in accuracy for imaging with gadodiamide and gadopentetate was 3.6%. Noninferiority was inferred because the lower bound of the exact two-sided 95% confidence interval was -10.1 and was above the noninferiority margin (-15%). Accuracy for detection of the main stenosis was low, 56.4% for gadodiamide and 52.8% for gadopentetate group. Subgroup analysis with exclusion of inferior mesenteric artery and internal iliac arteries and the most false-positive stenosis classifications yielded better results: 76.6% and 71.6%, respectively. Sensitivity, specificity, and negative and positive predictive values did not differ substantially between study groups. In the main analysis, values were 44%, 96%, 35%, and 97% for gadodiamide and 44%, 83%, 30%, and 90% for gadopentetate, respectively. In the subgroup analysis, values were 66%, 95%, 61%, and 96% for gadodiamide and 63%, 86%, 58%, and 88% for gadopentetate, respectively. Noninferiority of gadodiamide versus gadopentetate was verified based on the primary end point, which was accuracy for detection of the main stenosis with enhanced 3D MR angiography compared with DSA.