The purpose of this study was to assess the diagnostic value of two-dimensional (2D) MR subtraction angiography of lower extremities in patients with symptomatic peripheral arterial occlusive disease with conventional angiography as the standard of reference. Twenty patients were prospectively included. 2D subtraction MR angiography (MRA) consisted of multisection gradient-recalled echo (GRE) acquisitions with the shortest TE available on our machine (4 msec), obtained in the coronal plane before and after intravenous bolus administration of gadolinium chelate. MR images were reconstructed after subtraction with a maximum-pixel-intensity-projection (MIP) algorithm. MRA was performed in all cases 1-4 days before diagnostic angiography. In a prospective blinded analysis, the number and location of significant (ie, >50%) stenoses and occlusions were evaluated for each vascular segment. Sensitivity and specificity were used to evaluate MRA data. Significant stenoses (38 of 46, 83%) and occlusions (66 of 67, 99%) seen at conventional angiography were identified with MRA. The sensitivity and specificity of MRA for determination of stenoses >50% or occlusions was 100% and 97%, respectively. The location and extent of stenoses and/or occlusions on MRA and angiograms were well correlated (kappa values, r = .73, P < .05). Contrast 2D MR subtraction angiography, by providing comparable information to that of conventional angiography, is well suited to evaluate the presence and severity of atherosclerotic lesions of the lower limbs.